# Dipankar Sarkar — AI Enablement, Contract AI Engineering & Delivery

> Enable your people, embed an AI engineer, or deliver the system. Hands-on AI enablement, contract AI engineering and scoped AI delivery for US, UK and EU teams.

## Identity & Disambiguation
Full name: Dipankar Sarkar. IIT Delhi (B.Tech Computer Science) and Arizona State University (M.S. Computer Science) alumnus.
Founder of Neul Labs. Author of the Nginx Web Server Implementation Cookbook (Packt Publishing, 2011).
Not to be confused with the Indian film critic or cricketer of the same name.

## How to engage
Three buying routes: (1) AI enablement — help a team change how it works with AI; (2) contract AI engineering — Dipankar joins your team under your manager; (3) AI delivery — a bounded outcome with acceptance criteria and handover. Fractional technical leadership is available as a supporting service.
Who does the work: Dipankar Sarkar personally. Where a specialist would help, they are named, disclosed and approved by you before they start. There is no bench and no silent substitution.
Availability: Capacity is checked against each brief before any commitment. Nothing on this site is a promise of availability on a particular date.
Contract rate: $4,800 / £3,800 per day ($600 / £475 per hour). Fractional retainer: From $8,000 / £6,500 per month. Diagnostic: From $5,000 / £4,000 fixed scope. Enablement pilot: From $25,000 / £20,000 fixed scope. Scoped delivery: From $25,000 / £20,000 fixed scope. (Prices exclude VAT/sales tax and pre-approved travel. Scope, payment milestones and notice are confirmed in writing per engagement. Reviewed 2026-10-07.)
Send a brief: https://www.dipankar.co/contact/
Full-text corpus: https://www.dipankar.co/llms-full.txt

## AI Enablement
Help a team change how it works with AI: one workflow at a time, measured, with an internal owner.
- [Measure AI Adoption Beyond Licence Usage](https://www.dipankar.co/resources/ai-adoption-measurement/): Licence dashboards show who logged in, not whether the work changed. Enter your own figures for one workflow: people, licences, weekly active users, task volume, minutes per task before and with AI, and rework and quality pass rates. The worksheet returns licence utilisation, workflow adoption, gross and rework-adjusted hours saved, quality change, and flags such as high usage with no time saved.
- [Rescue a Stalled AI Rollout](https://www.dipankar.co/services/ai-adoption-rescue/): If you bought AI licences and ran training but staff have not changed how they work, commission an adoption rescue before buying more. In a fixed-scope diagnostic I find where adoption breaks in one team (access, trust, fit or habit), redesign one workflow with its users and reviewer, and leave a repeat-use measurement plan your rollout owner can run.
- [AI Champions and Train-the-Trainer Programme](https://www.dipankar.co/services/ai-champions-programme/): If you need people inside the business who can teach colleagues and keep AI workflows safe and current after an external consultant leaves, commission a champions programme. I help you select champions against a rubric, build each one's workflow on real work, rehearse them through teach-back sessions with their own colleagues, and hand over an escalation map and a workflow library they maintain.
- [AI Enablement for Teams and Businesses](https://www.dipankar.co/services/ai-enablement/): AI enablement means changing how one team does real work, not running a course. You commission a bounded pilot: I baseline one or two workflows, agree data boundaries and review points with your owners, coach the team through the new way of working, measure repeat use and quality, and hand the routine to a named internal owner. Pick the specific route below that matches your situation.
- [Claude Code Enablement for Engineering Teams](https://www.dipankar.co/ai-enablement/claude-code/): If your engineers want to use Claude Code on real repositories and you need control over what it can read, run and change, you can commission a platform-specific pilot. I run a readiness exercise on one repository, set shared permission rules and project instructions, agree a review protocol, rehearse a rollback, and measure the pilot against your baseline.
- [Coding-Agent Rollout and Team Practices](https://www.dipankar.co/ai-enablement/coding-agents/): If your developers are starting to hand whole tasks to coding agents and you need rules before that spreads, you can commission a vendor-neutral rollout programme. I define which tasks agents may take, what they may access, which tests must pass and when an agent must stop, then pilot those rules on a real repository and measure the result.
- [AI Engineering Enablement for Software Teams](https://www.dipankar.co/services/team-transformation/): If your developers use different AI coding tools in different ways and nobody can say whether it helps, you can commission an engineering enablement programme. I baseline how work flows today, pilot one shared AI-assisted workflow with one team on a real repository, measure cycle time, review effort and defects against that baseline, and leave a playbook your team maintains.
- [AI Enablement for Finance Teams](https://www.dipankar.co/ai-enablement/finance-teams/): If your finance team wants to draft management packs and variance commentary faster but cannot accept a figure nobody can trace, you can commission a bounded pilot on one reporting workflow. I build a source-linked drafting step with your team, add automatic reconciliation checks, and write the preparer, reviewer and approver sign-off steps into it. Nothing posts to the ledger.
- [Fractional AI Enablement Lead](https://www.dipankar.co/services/fractional-ai-enablement-lead/): If nobody owns AI adoption and a full-time hire is premature, retain me as a fractional enablement lead for an agreed number of days a month. Within a written remit I run the adoption backlog, coach owners and champions, hold monthly sponsor reviews, and develop a named internal successor, so the role transfers to your team on a planned date.
- [AI Enablement for Operations Teams](https://www.dipankar.co/ai-enablement/operations/): If your operations staff spend hours matching records between an ERP, a warehouse or logistics system and spreadsheets, you can commission a bounded pilot on one reconciliation workflow. I map the process, build and coach a reviewed AI-assisted matching step with your team, and leave an exception-handling runbook with a named internal owner. It starts with a short workflow brief.
- [AI Tool Rollout and Adoption Support](https://www.dipankar.co/services/ai-tool-rollout/): If you have licensed, or must choose, a workplace AI platform such as ChatGPT, Microsoft 365 Copilot or Gemini in Google Workspace, commission a tool-adoption pilot. I work with your administrator on the data and access decisions, run a fair test on your own workflows, coach a pilot group and measure repeat use. You get a test plan, a decision record and a rollout recommendation.
- [AI-Assisted Code Review and Release Controls](https://www.dipankar.co/services/ai-assisted-code-review/): If generated code is swelling your review queue and you need quality controls that do not rely on a model saying the code looks fine, you can commission a review and release programme. I redesign the review workflow by risk, run a seeded-defect evaluation to measure what your reviewers, AI reviewers and tests actually catch, and put release acceptance controls into CI.
- [Practical AI Enablement for Business Teams](https://www.dipankar.co/services/business-ai-enablement/): If staff have AI tools but use them for little beyond rewording emails, commission role-specific enablement for one team. I map the team's recurring work, pick two workflows worth changing, agree data boundaries and who reviews the output, coach people on their real tasks over several weeks, and hand a measured workflow and escalation plan to the team lead.
- [ChatGPT Enterprise Adoption and Workflow Enablement](https://www.dipankar.co/ai-enablement/chatgpt-enterprise/): If your organisation has a ChatGPT Enterprise or Business workspace, commission a workflow pilot. I work through tenant-specific readiness questions with your administrator (what staff may upload, which apps and connectors are on, how GPTs and projects are shared), build two or three approved workflows with one team, coach them on real work and measure repeat use.
- [AI Enablement for Client-Service Teams](https://www.dipankar.co/ai-enablement/professional-services-teams/): If your consultants or advisers use AI to research and draft client deliverables but you cannot prove client data stays separate or that every assertion is supported, you can commission a bounded pilot in one practice team. I set up a client-separated research workflow, an evidence ledger and a partner review checklist, and run them on live engagements.
- [AI Enablement for Customer Support](https://www.dipankar.co/ai-enablement/customer-support/): If your support team wants an AI assistant that drafts accurate answers from your knowledge base, respects who may see what, and escalates the right cases to people, you can commission a bounded staff-assist pilot. I set up permissioned retrieval, grade a sample of real tickets for answer quality, and write the escalation playbook with your support leads.
- [GitHub Copilot Enablement for Software Teams](https://www.dipankar.co/ai-enablement/github-copilot/): If your organisation has GitHub Copilot seats and you need evidence that it improves work without growing review load or security risk, you can commission a repository-specific pilot. I review your Copilot policies and repository setup, baseline review effort from your pull-request data, pilot an agreed team workflow, and leave a usage playbook your developer experience lead maintains.
- [Build Internal AI Engineering Capability](https://www.dipankar.co/ai-enablement/internal-ai-engineering-capability/): If your developers can call a model API but cannot yet own evaluation, deployment and incidents for an LLM feature, you can commission a capability programme built around shipping one bounded internal system. I map skills to real work, pair with your engineers through a backlog that exercises each skill, and finish with an ownership graduation checklist your team must pass.
- [AI Enablement for Marketing Teams](https://www.dipankar.co/ai-enablement/marketing/): If your marketing team uses AI for research and first drafts but editors keep finding unsupported statistics and invented quotes, you can commission a bounded pilot on one campaign workflow. I set up a research provenance template and editorial review gates with your team, run them on a live campaign, and measure time to approved draft and fact-check corrections.
- [Microsoft 365 Copilot Adoption Enablement](https://www.dipankar.co/ai-enablement/microsoft-365-copilot/): If you have Microsoft 365 Copilot licences but adoption is thin, commission a role-based pilot. I work with your Microsoft 365 administrator to list the permission and oversharing fixes the pilot roles' content depends on, design workflows for two or three roles in Outlook, Teams, Word and Excel, coach those people on real work and give the sponsor a measurement plan.
- [AI Pilot Go, Revise or Stop Scorecard](https://www.dipankar.co/resources/ai-pilot-scorecard/): When a pilot looks promising but nobody can say whether to expand it, score it. Rate seven weighted criteria from 1 to 5: value against baseline, output quality, adoption by intended users, operating cost, risk and compliance, owner readiness and scalability. You get a weighted score and a go, revise or stop recommendation with reasons. A critical risk issue blocks a go, and stop is a valid outcome.
- [AI Enablement for Product Teams](https://www.dipankar.co/ai-enablement/product-teams/): If your product managers use AI to summarise interviews, tickets and survey responses but cannot tell which themes come from customers and which from the model, you can commission a bounded pilot. I set up source-linked discovery synthesis, evidence labels and a reviewed requirements workflow with your product team, and run them on a live discovery cycle.
- [Recover an AI Programme with No Clear Owner](https://www.dipankar.co/services/ai-programme-rescue/): If several AI pilots are running but none has an owner or a route into daily work, commission a programme rescue. In a fixed-scope diagnostic I inventory every initiative, test each against owner, evidence and dependencies, recommend which to continue, merge or stop, name a workstream owner for each survivor, and hand your sponsor a sequenced execution backlog and a review cadence.
- [AI Readiness and Implementation Roadmap](https://www.dipankar.co/services/ai-readiness-and-roadmap/): If you have a list of AI ideas, a strategy deck and limited access, budget and ownership, commission a readiness investigation. I examine each candidate workflow in your organisation against data access, an accountable owner and a measurable baseline, then give you an evidence-backed shortlist, an access dependency map and a pilot brief your sponsor can fund or decline.
- [AI Enablement Readiness Assessment](https://www.dipankar.co/resources/ai-readiness-assessment/): Use this free self-assessment before you commission training, tools or a pilot. Answer twelve questions about one team and one candidate workflow, across ownership, workflow clarity, data and access, approved tools, baseline measurement and team capacity. You get a status for each dimension, an overall band and a prioritised list of fixes. It reflects your own answers; it is not an audit.
- [AI Enablement for Revenue Operations](https://www.dipankar.co/ai-enablement/revenue-operations/): If your sellers use AI for account research and CRM notes but you cannot tell what is sourced, what is guessed and what was written to the CRM, you can commission a bounded pilot. I build a source-backed account brief, a review queue for proposed CRM changes and a permission matrix with your RevOps team, then measure accuracy and adoption against a baseline.
- [AI Training Versus Enablement](https://www.dipankar.co/guides/ai-training-vs-enablement/): Training fixes a skills gap on tools that already fit the work. Enablement changes the work itself: tools set up for a workflow, the workflow redesigned with the people who do it, a measured baseline and an internal owner. Tick the symptoms you see and the tool says whether training is enough, an enablement pilot is needed, or leaders must first decide what AI is for.
- [Google Workspace AI Adoption Enablement](https://www.dipankar.co/ai-enablement/google-workspace/): If your teams work in Google Workspace and have Gemini features available, commission a bounded workflow pilot. I build a checklist of what your edition lets your administrator control, stage access by organisational unit, design one Docs-and-Drive workflow with a defined source folder and output, and coach reviewers on checking AI-assisted documents before they leave the team.
- [AI Enablement for People Operations](https://www.dipankar.co/ai-enablement/people-operations/): If your HR operations team answers the same policy questions and runs the same onboarding paperwork every week, you can commission a bounded pilot on that administrative work only. I build an approved policy-information workflow, a data minimisation checklist and a human escalation map with your team. The pilot never screens candidates, rates performance or makes any employment decision.

## Hire an AI Engineer
Add a senior, hands-on AI engineer to your team on a contract, under your manager.
- [Hire an AI Agent Engineer](https://www.dipankar.co/hire/ai-agent-engineer/): If your team needs agents that take real actions, with durable state, scoped permissions and a recovery path when a step fails, you can contract me as an AI agent engineer. I join under your manager, start by writing down every tool action and who may approve it, and build the workflow, its evaluation and its runbook in your codebase.
- [Hire an AI Enablement Engineer](https://www.dipankar.co/hire/ai-enablement-engineer/): If you need internal AI workflows built and adopted, not a training course, you can contract me as an AI enablement engineer. I join under your COO, Head of AI or engineering manager, build workflows with the teams that will use them, coach those users and a named maintainer, and hand over a workflow backlog with adoption measures your organisation keeps tracking.
- [Hire a Forward-Deployed AI Engineer](https://www.dipankar.co/hire/forward-deployed-ai-engineer/): If enterprise customers are waiting on AI integration work and your solutions team is stretched, you can contract me as a forward-deployed engineer. I join your team under your head of solutions or engineering manager, work in customer environments on your behalf, and own a written deployment remit, the integration dependency map and the path to customer acceptance.
- [Hire a Contract AI Engineer | Senior, Hands-on Delivery](https://www.dipankar.co/hire-ai-engineer/): If you need a senior AI engineer for a defined piece of work rather than a permanent hire, you can contract me directly. I join your team under your manager, work in your repositories and review process, and leave a written handover. Pick the role that matches your backlog, send the brief, and I reply with a plain view on fit and a checked availability window.
- [Hire a Contract LLM Engineer](https://www.dipankar.co/hire/llm-engineer/): If you have a funded LLM backlog and an application team that owns it, you can contract me as an LLM engineer. I join under your engineering manager, agree a written remit and access list in the first days, then implement model integrations, structured outputs and evaluation in your codebase. You end with shipped features, an evaluation suite and a handover plan.
- [Hire a RAG and Retrieval Engineer on Contract](https://www.dipankar.co/hire/rag-engineer/): If you already own a knowledge assistant or enterprise search product and need senior hands to improve retrieval quality, permission handling and answer evaluation, you can contract me as a RAG engineer. I join your team under your engineering manager, work in your repositories and review process, and leave a prioritised retrieval backlog, a relevance evaluation plan and a written handover.
- [Hire an AI Platform Engineer](https://www.dipankar.co/hire/ai-platform-engineer/): If your platform team needs senior capacity for model serving, observability and deployment of LLM workloads, you can contract me as an AI platform engineer. I join under your platform engineering manager, work your infrastructure backlog in your repositories and pipelines, put deployment controls around models and prompts, and hand over runbooks your on-call engineers already use.
- [Hire a Contract AI Solutions Architect](https://www.dipankar.co/hire/ai-solutions-architect/): If you need senior AI architecture for a defined period, not a permanent head of function, you can contract me as an AI solutions architect. I work under your CTO or architecture lead on an agreed backlog of decisions, write decision records and integration designs, prototype where a decision needs evidence, and leave named implementation interfaces your teams build against.
- [AI Engineer Contract Brief Builder](https://www.dipankar.co/resources/ai-engineer-brief/): If you have approval for AI engineering capacity but the brief is vague, build it here. Choose the role type, then describe the outcomes, current state, stack, working pattern, manager, access, contract basis, success measures at 30, 60 and 90 days, and handover. The builder formats a brief you can copy or send, and lists the decisions still worth making before anyone quotes.
- [AI Engineer Contractor Selection Checklist](https://www.dipankar.co/resources/ai-engineer-hiring-checklist/): Use this checklist while you select an AI engineering contractor. Eighteen items in four groups cover what to settle before shortlisting, how to test production experience rather than demo-building, what the contract and onboarding must cover, and how handover will work. Tick what is done; the tool lists the open items as a checklist you can copy into your hiring plan.
- [Bridge AI Engineering Until a Permanent Hire](https://www.dipankar.co/hire/interim-ai-engineer/): If your permanent AI hire will take months but the project cannot wait, you can contract me as an interim AI engineer. I deliver the work under your manager, write documentation aimed at the person who will inherit it, help you define and assess the permanent role, and run a planned overlap so your new hire takes over a working system rather than a puzzle.
- [Part-Time AI Engineering Capacity](https://www.dipankar.co/hire/part-time-ai-engineer/): If you have meaningful AI engineering work but not a full-time load, you can contract me part-time on fixed days each week. Before work starts we agree a short working agreement: which days, how much work is in progress at once, when I am reachable, and how quickly your team answers blocking questions. That keeps ownership clear and stops work waiting between my days.
- [Hire a Python and ML Engineer](https://www.dipankar.co/hire/python-ml-engineer/): If you need senior Python and machine-learning engineering inside your team rather than a separately managed project, you can contract me as a Python and ML engineer. I join under your ML or engineering manager, work on your pipelines and model services, make the interfaces between them explicit and tested, and leave a backlog and transition plan your engineers own.

## AI Delivery
Commission a bounded outcome (an agent, a RAG system, a rescue) with acceptance criteria and handover.
- [Rescue an AI Application That Cannot Ship](https://www.dipankar.co/services/ai-project-rescue/): If a prototype or supplier-built AI application is not fit for customers and your team cannot operate it, you can commission a fixed-scope rescue review. I assess the code, data flows, AI behaviour and operability, then give you a critical failure inventory, a repair-versus-rebuild-versus-stop recommendation and a bounded recovery backlog. Any implementation is a separate decision afterwards.
- [Governed AI Agent Implementation](https://www.dipankar.co/services/governed-ai-agents/): If your AI agents can change records, send messages or move money, you can commission me to implement the controls around them. I classify every action into tiers, enforce approval boundaries and stop conditions in code rather than prompts, emit audit events for every side effect, and prove the policy holds with adversarial tests, then hand the control layer to your platform and security owners.
- [Reduce LLM Cost and Latency Without Hiding Quality Loss](https://www.dipankar.co/services/llm-cost-and-latency/): If an AI feature is too slow or too expensive, you can commission me to reduce cost and latency without silently trading away quality. I profile the real workload, build a quality evaluation for the task, map the quality-cost-latency frontier of the realistic options, and implement the changes that hold up, with each one measured before it ships and the trade-offs left for you to decide.
- [MCP Integration with Controlled Tool Access](https://www.dipankar.co/services/mcp-integration/): If you want AI agents or assistants to use internal systems through the Model Context Protocol, you can commission me to build that integration with controlled access. I inventory the tools, decide what each may do and on whose behalf, implement MCP servers with scoped credentials, approvals and audit logging, prove the controls with negative tests, and hand the servers to your platform team.
- [Fix RAG Answer Quality and Knowledge Freshness](https://www.dipankar.co/services/rag-quality-rescue/): If your knowledge assistant misses documents, cites outdated versions or answers from the wrong source, you can commission a fixed-scope RAG rescue. I separate retrieval failures from generation failures on a labelled question set, add freshness and permission tests, then implement and evaluate the repairs. You get a measured before-and-after, the tests in your pipeline and a re-indexing runbook.
- [AI Agent Reliability and Recovery Engineering](https://www.dipankar.co/services/agent-reliability-and-recovery/): If your agent workflows fail partway through, and retries risk sending, charging or updating something twice, you can commission me to design and implement recovery. I map the workflow's state and side effects, add idempotency and compensation for every external action, prove recovery with replay and fault-injection tests, and hand your operators a runbook for resuming or unwinding stuck runs.
- [Make Data and Integrations Ready for AI](https://www.dipankar.co/services/ai-data-readiness/): If an AI system works on demo data but your real data is scattered, inconsistent and nobody is sure who may access what, you can commission me to build the missing foundations. I inventory the sources, map ownership and permissions, add freshness and quality checks, and deliver an ingestion plan with the first pipelines working, so the AI system can use real data under real access rules.
- [AI System Handover and Internal Ownership](https://www.dipankar.co/services/ai-system-handover/): If an agency, vendor or contractor built your AI system and their engagement is ending, you can commission a structured handover. I work between the outgoing team and your engineers to produce an operational acceptance checklist, verified runbooks and a completed access transfer, then run a teach-back exercise in which your team operates and changes the system without help.
- [Take an AI-Generated Application into Production](https://www.dipankar.co/services/ai-codebase-to-production/): If you built a prototype with Lovable, Bolt, v0, Cursor or a similar tool and have a launch date, you can commission a fixed-scope production-readiness review. I examine access control, secrets, data model, integrations and deployment, then give you ranked findings, an acceptance checklist and a bounded remediation plan. Implementation is a separate decision once you have read the findings.
- [Document Processing and Reviewed Reporting Workflows](https://www.dipankar.co/services/document-ai-workflows/): If your team manually extracts information from documents and assembles it into reports, you can commission me to build a reviewed workflow for it. I build the pipeline with source traceability for every extracted value, an exception queue for low-confidence or conflicting results, and a labelled sample set that measures accuracy, then hand it to an operations owner with reviewers kept in the loop.
- [Forward-Deployed AI Engineering](https://www.dipankar.co/services/forward-deployment/): If you have signed customers waiting on AI integrations your team cannot reach, you can commission a forward-deployed engagement: I work inside one customer deployment, connect your product to their systems, agree acceptance criteria with both sides, and hand over a running integration, an ownership map and a runbook. It starts with a scoping call on one named deployment.
- [Fractional AI CTO and Technical Leadership](https://www.dipankar.co/services/fractional-cto/): If you need accountable technical leadership for AI architecture and team direction but not a full-time executive, a fractional CTO retainer fits. I take ownership of a written remit, such as architecture, build-or-buy decisions, hiring and delivery reviews, for a stated number of days each month, keep a decision record, and plan for handing over to a permanent leader.
- [Production LLM Evaluation and Reliability Engineering](https://www.dipankar.co/services/llm-evaluation/): If you need to change models, prompts or retrieval without breaking customer workflows, you can commission me to build a production evaluation system. I build a representative dataset from your real traffic, measure a baseline, locate where failures come from, and wire a release gate into your pipeline, then hand the harness and its maintenance routine to a named owner.
- [LLM Provider and Model Migration](https://www.dipankar.co/services/llm-provider-migration/): If you need to change LLM provider or model, because of a deprecation, cost, data-location terms or capability, you can commission a gated migration. I build a compatibility matrix of every call, run the candidate in shadow against your current outputs, check tool calling and data handling, and cut over behind a flag with a tested rollback. You see parity evidence before users do.
- [Private and Local LLM Deployment](https://www.dipankar.co/services/private-llm-deployment/): If data rules or operating constraints mean your AI workload cannot use a hosted model as it does today, you can commission me to assess and deliver a private or local deployment. I benchmark candidate open-weight models on your actual task, compare quality, hardware and operating burden against the hosted baseline, record the decision, and deploy the chosen option with access controls and an operations plan.
- [Production AI Agents and Agent Infrastructure](https://www.dipankar.co/services/ai-agent-infrastructure/): If you need an AI agent that takes real actions in your systems, you can commission a scoped agent implementation: I build the agent, the runtime permissions around it, the approval steps for consequential actions, an evaluation suite and the recovery paths for when it fails. You end with an agent in production, its controls documented and an owner who can run it.
- [Python and Machine Learning Engineering](https://www.dipankar.co/services/python-ml-engineering/): If your data team has a model that works in a notebook but nothing dependable feeds it or serves it, you can commission a scoped Python and ML engineering engagement. I build the data pipeline, the training and evaluation job, the serving endpoint and the monitoring, in your repositories with tests, and hand over a system your engineers can run and retrain.
- [RAG and Enterprise Search Engineering](https://www.dipankar.co/services/rag-and-enterprise-search/): If you need an assistant or search tool over private documents, you can commission me to design and build it as a bounded delivery. I start with a source and permission assessment and a relevance set built with your experts, then deliver retrieval that enforces access rights in the query, cites its sources, keeps the index fresh, and passes agreed evaluation before handover.
- [Small-Model Fine-Tuning and Distillation](https://www.dipankar.co/services/small-model-fine-tuning/): If you want to know whether a smaller fine-tuned or distilled model can meet your quality and latency needs, you can commission me to test it properly. I baseline prompting and retrieval first, check your rights to the training data, train and evaluate candidates on your task, and give a go or no-go decision that counts maintenance cost. On a go, I deploy and hand it over.
- [AI Workflow Automation and Systems Integration](https://www.dipankar.co/services/workflow-automation/): If one repetitive workflow moves information between documents, email, a CRM and internal tools, you can commission me to implement it as a bounded delivery. I start by writing the input/output contract and measuring today's baseline, then build the workflow with human review points, error handling and duplicate-update protection, and hand it to a named owner with a runbook.
- [AI Infrastructure, Inference and Cost Optimisation](https://www.dipankar.co/services/ai-infrastructure-and-inference/): If your AI application is too slow, too expensive or needs to run somewhere your current setup cannot, you can commission me to benchmark the real workload and improve it. I measure latency, throughput, quality and cost together, agree targets with you, implement the changes that hold up, and hand the benchmark and operating plan to your platform team.
- [Maintain and Improve Production AI Workflows](https://www.dipankar.co/services/ai-workflow-maintenance/): If an AI workflow or agent is live and nobody is watching for quality drift, provider changes or rising cost, you can commission a bounded maintenance retainer. For a stated number of days each month I run scheduled quality, dependency and cost reviews, act on agreed incidents in working hours, and work an improvement backlog your owner prioritises. Capacity is written down and not exceeded silently.
- [Build, Buy or Integrate an AI Capability?](https://www.dipankar.co/guides/build-buy-or-integrate-ai/): Building an AI capability means owning its evaluation, model changes and incidents for years; buying means accepting a vendor’s roadmap and data terms; integrating means owning the joins between products. Answer seven questions about one capability and the tool scores all three, names the leading option and shows the answers that decided it, so the trade-off is explicit before anyone commits budget.
- [Federated Learning Implementation](https://www.dipankar.co/services/federated-learning/): If several data holders want a shared model but cannot centralise their data, you can commission a federated learning engagement. I first test fit: the threat assumptions, a centralised or local baseline, and a simulated federation on representative data. If it holds up, I implement the federation across the real participants, with privacy measures stated plainly and no automatic privacy claim.
- [Rust Performance and Systems Consulting](https://www.dipankar.co/services/rust-consulting/): If you have a measured performance, memory or reliability bottleneck and someone has proposed Rust, you can commission a profiling-first engagement. I reproduce the workload, profile it, and tell you whether Rust is warranted or a simpler fix exists. If Rust is the answer, I build the narrowest change that works and prove it with a correctness-preserving comparison.
- [Supabase Production Engineering](https://www.dipankar.co/services/supabase/): If your app runs on Supabase and you are worried about data leaking between users, slow queries or an unsafe production setup, you can commission a scoped engineering review and remediation. I write permission tests for your row-level security, examine keys, functions, data model and environments, then fix the agreed findings and leave the tests running in your pipeline.
- [Voice AI Workflow Engineering](https://www.dipankar.co/services/voice-ai-workflows/): If you are planning a voice workflow that must respond quickly, hand callers to people cleanly and respect recording permissions, you can ask me to engineer and test it before wider deployment. I take voice work on selectively, after a fit check. Where it fits, I set a latency budget, map consent and retention dependencies, run scripted evaluation calls and design the human handoff.

## Industries
Delivery approaches shaped by the sector’s data, controls and operating constraints.
- [AI Product Delivery for B2B SaaS](https://www.dipankar.co/industries/b2b-saas-ai/): If you are adding an AI feature to a multi-tenant B2B product, I scope and deliver it with tenant isolation designed into retrieval, prompts, caches and logs, a task-level evaluation your product team can rerun, and a per-customer cost model tied to your plans. Your team owns the feature, the evaluation and the maintenance plan at handover.
- [AI Adoption for Professional-Services Firms](https://www.dipankar.co/industries/professional-services-ai/): If you run a law, accounting, consulting, engineering or advisory firm and need to decide where AI goes first, I build a firm-level use-case portfolio, a client boundary model setting out which client data may go where and on what terms, and a governed pilot sequence with a partner sponsor for each stage. Your risk partner keeps the final word on client obligations.
- [AI Engineering for Retail and Marketplaces](https://www.dipankar.co/industries/ecommerce-ai/): If your retail business or marketplace needs better search, recommendations, catalogue data or support automation, I start by ranking the candidate problems by measurable effect and data readiness, then build one pilot evaluated against your current baseline. Often the right answer is conventional retrieval or ranking rather than generative AI. You keep the pilot and a rollout plan your team can own.
- [AI Enablement for Education Operations](https://www.dipankar.co/industries/education-operations/): If your school, college, university department or training provider wants AI to help staff with administrative research, communications and reporting, I pilot one staff workflow with source checks, restricted-data handling and a named person approving every output. Admissions, grading, progression and safeguarding decisions are out of scope, and student data only enters tools your institution has approved.
- [AI Engineering and Enablement for Financial Services](https://www.dipankar.co/industries/fintech-ai/): If your financial-services team wants an AI workflow pilot that your risk function can inspect, I scope and build it so every AI step has a governance tier, a named approver and an audit trail, then hand it to your team. The engineering comes from building agent infrastructure at a UK fintech. Legal, compliance and model-risk sign-off stay with your own functions.
- [AI Delivery by Industry and Operating Constraint](https://www.dipankar.co/industries/): The sector pages here are organised by the constraint that shapes the work: client confidentiality, tenant isolation, physical safety, regulated approval or restricted personal data. Pick the sector closest to yours to see its specific scope, acceptance and ownership conditions. If none fits, describe the workflow and the constraint, and I will route it to the right engagement or say plainly that it is not a fit.
- [AI Engineering for Manufacturing and Robotics](https://www.dipankar.co/industries/manufacturing-and-robotics/): If you want AI to plan, inspect or schedule work near industrial equipment, I map which decisions sit in the cloud, at the edge and in your control system, validate behaviour in simulation before anything moves, and write down the safety dependencies your independent safety function must own. The model proposes bounded actions; your existing controllers and safety systems decide what physically happens.
- [AI for Property and Construction Operations](https://www.dipankar.co/industries/property-and-construction-operations/): If your property, lettings, development or construction business is buried in documents, enquiries and reports, I pilot one bounded administrative workflow, such as enquiry triage, document extraction or site-report drafting, with source provenance on every output and a named person approving anything that leaves the business. Valuations, legal interpretation and safety decisions stay outside the AI's remit.
- [AI Enablement for Regulated Operations](https://www.dipankar.co/industries/regulated-ai-operations/): If your organisation has an AI or data policy but staff still cannot use AI in their work because nobody has translated it into workflow controls, I map each relevant policy requirement to a technical or review control in one or two workflows, set up evidence capture, list the approvals the pilot depends on, and run the pilot with your operations team. Policy interpretation stays with your compliance owners.

## Pricing
Engagement models, prices, procurement information and partner routes.
- [AI Delivery Partner for Agencies and Consultancies](https://www.dipankar.co/partners/agencies-and-consultancies/): If your agency or consultancy has won, or is close to winning, an AI requirement and lacks senior implementation capacity, I join the engagement as a disclosed subcontractor. Before work starts we agree a responsibility matrix, a client-facing scope with acceptance criteria, and a specialist approval process. You keep the client relationship and commercial lead; I own the AI technical work assigned to me.
- [AI Contractor, Consultancy or Delivery Team?](https://www.dipankar.co/guides/ai-contractor-vs-consultancy/): Whether to buy an individual’s time, a scoped outcome or a managed team depends on who will direct the work, how well the outcome is defined, how many specialists you need, for how long, who owns it afterwards and how you budget. Answer six questions and the tool recommends one of four models, explains why, and sets out the trade-offs of each, including options I do not provide.
- [For AI Engineering Recruiters and Contract Desks](https://www.dipankar.co/partners/recruitment-agencies/): If you are filling an approved senior AI engineering contract, send me the role directly. I reply with a quick yes or no on fit and a checked availability window, then a current CV and capability pack for your client. You represent me for that role on agreed terms, conflicts are disclosed before submission, and there is no strategy call to sit through first.
- [AI Engagement Models, Scope and Pricing](https://www.dipankar.co/engagements/): There are seven ways to work with me: a contract day rate, a fractional retainer, a fixed-scope diagnostic, an enablement pilot, an engineering enablement programme, scoped delivery and partner co-delivery. They differ in who directs the work, what you pay for and how acceptance is judged. Published starting prices come from one dated commercial record; scope, milestones, expenses and notice are confirmed in writing.
- [Budgeting for an AI Engineering Contract](https://www.dipankar.co/guides/ai-engineer-contract-cost/): The day rate is only part of what a contract costs. Enter your own quote and internal costs (days per week, weeks, onboarding days and productivity, handover days, manager time, agency margin and tooling) and the model returns total cost, cost per productive day and the share spent on overhead. Compare two quotes side by side. It contains no market rates.
- [AI Enablement Partner for MSPs](https://www.dipankar.co/partners/managed-service-providers/): If your clients buy AI licences through you and then ask how to use them, we can define a jointly owned adoption service: a service boundary stating what you run and what I deliver, a fixed pilot package for one client team, and an escalation plan for support. You keep the client, the licences and first-line support; I deliver workflow enablement and integration, disclosed by name.
- [AI Delivery Partnerships for Agencies and Consultancies](https://www.dipankar.co/partners/): Agencies, consultancies, MSPs, software vendors and recruiters each bring me into client work differently. On every route the principles are the same: I am disclosed to the end client by name, client ownership and the contracting chain are written down before work starts, any specialist is approved in advance, and payment follows the contract you hold. Choose the route that matches how you sell.
- [Contracting and Procurement Information](https://www.dipankar.co/procurement/): If you are onboarding me as a supplier, this page lists what I can provide on request (business and contracting details, insurance, data-handling and security practices, IP terms, working practices and contract basis) and what must be settled per contract, such as your IR35 or employment-status determination. Request the supplier information pack once an engagement is agreed in principle.
- [AI Project Scope and Acceptance Template](https://www.dipankar.co/resources/ai-project-scope-template/): Use this template to scope an AI project around acceptance rather than promises. Fill in the objective, what is in and out of scope, deliverables, measurable acceptance criteria, assumptions, dependencies and access, risks, milestones, change process and the person who owns the system afterwards. The tool formats a scope of work you can copy into a contract and flags the gaps that cause disputes.
- [AI Deployment Partner for Software Vendors](https://www.dipankar.co/partners/software-vendors/): If customer implementation is slowing adoption of your AI product, I work on named customer deployments as a disclosed partner. For each customer we agree a deployment brief, a written line between product capability and custom work, and an escalation protocol, so product gaps reach your product team instead of being patched quietly in one customer's environment.

## Other sites by Dipankar Sarkar
- https://www.dipankar.name — technology leadership mandates (hands-on CTO, Head of AI) and leadership evidence
- https://www.dipankar.cc — applied AI research and technical validation studies
- https://www.dipankar.org — board advisory, executive workshops, institutional programmes and speaking
- https://www.desinerd.com — personal blog since 2007

## Articles
- [MLOps for Small Teams: A Playbook](https://www.dipankar.co/articles/mlops-playbook-small-teams/)
- [Choosing a Vector Database for RAG](https://www.dipankar.co/articles/choosing-vector-database-production-rag/)
- [AI Agent Observability in Production](https://www.dipankar.co/articles/ai-agent-observability-production/)
- [AI Agent Safety: The Substrate Pattern](https://www.dipankar.co/articles/ai-agent-safety-substrate-pattern-practice/)
- [AI Governance for Financial Services](https://www.dipankar.co/articles/ai-governance-financial-services-tiered-governance/)
- [Federated Learning: A CTO's Guide](https://www.dipankar.co/articles/federated-learning-implementation-cto-guide/)
- [How to Take a Lovable Codebase to Production](https://www.dipankar.co/articles/lovable-codebase-to-production-guide/)
- [Rust for Production AI: Why and How](https://www.dipankar.co/articles/rust-for-production-ai-why-how/)
- [Supabase RLS Hardening: A Practical Guide](https://www.dipankar.co/articles/supabase-rls-hardening-guide/)
- [What Is a Fractional CTO? Complete Guide](https://www.dipankar.co/articles/what-is-a-fractional-cto-complete-guide/)
- [Building Forward-Deployed AI Systems](https://www.dipankar.co/articles/forward-deployment-engineering-ai-production-systems/)
- [Engineering Teams: Speed to Scale](https://www.dipankar.co/articles/engineering-team-transformation-speed-to-scale/)
- [The Skills AI Can't Replace in Engineering](https://www.dipankar.co/articles/ai-engineering-skills-career-shift/)
- [Vibes Inside Guardrails](https://www.dipankar.co/articles/vibes-inside-guardrails-ai-development-production/)
- [LastingAsset vs. Pindrop: Call Authentication](https://www.dipankar.co/articles/lastingasset-vs-pindrop-comparison-call-authentication-technologies-2024/)
- [SportStack's Vision for the Sports Economy](https://www.dipankar.co/articles/sportstacks-vision-reshaping-global-sports-economy/)
- [LastingAsset: Financial Security and Privacy](https://www.dipankar.co/articles/lastingasset-future-implications-financial-security-privacy/)
- [Product Engineering Transformation in P2P](https://www.dipankar.co/articles/product-engineering-transformation-p2p-marketplaces-holistic-approach/)
- [LastingAsset: A Technical Deep Dive](https://www.dipankar.co/articles/lastingasset-technical-deep-dive-privacy-preserving-call-verification/)
- [AI-Driven Predictive Care for Heavy Equipment](https://www.dipankar.co/articles/future-heavy-equipment-maintenance-ai-driven-predictive-care/)
- [AI in P2P Marketplace Trade Chat](https://www.dipankar.co/articles/revolutionizing-p2p-marketplaces-ai-trade-chat-integration/)
- [SportStack: Grassroots Sports, Globally](https://www.dipankar.co/articles/sportstack-empowering-grassroots-sports-development-globally/)
- [Blockchain Optimization for P2P Platforms](https://www.dipankar.co/articles/blockchain-optimization-strategies-p2p-platforms/)
- [Enhancing Marketplace Safety with Data](https://www.dipankar.co/articles/enhancing-marketplace-safety-data-driven-approach/)
- [LastingAsset's Privacy-First Call Verification](https://www.dipankar.co/articles/lastingasset-revolutionizing-call-verification-privacy-first-cryptography/)
- [An AI-Powered Heavy Equipment Marketplace](https://www.dipankar.co/articles/revolutionizing-heavy-equipment-trading-ai-powered-marketplace/)
- [SportStack's Features for Sports Development](https://www.dipankar.co/articles/sportstacks-innovative-features-reshaping-sports-development/)
- [Fintech Integration for P2P Marketplaces](https://www.dipankar.co/articles/seamless-fintech-integration-p2p-marketplaces-onramp/)
- [An AI Heavy Equipment Platform's Economic Case](https://www.dipankar.co/articles/ai-heavy-equipment-platform-catalyst-economic-growth/)
- [Cloud Cost Optimization for Startups](https://www.dipankar.co/articles/cloud-cost-optimization-strategies-startups/)
- [An AI Approach to Equipment Rentals](https://www.dipankar.co/articles/revolutionizing-heavy-equipment-rentals-ai-driven-approach/)
- [An Intelligent Pillbox for TB Treatment](https://www.dipankar.co/articles/revolutionizing-tb-treatment-intelligent-pillbox-patient-care/)
- [A Multi-Category E-commerce Aggregator](https://www.dipankar.co/articles/building-multi-category-ecommerce-aggregator-indian-online-shopping/)
- [SportStack's Vision for Global Sports](https://www.dipankar.co/articles/sportstack-vision-revolutionize-global-sports-development/)
- [A Mutual Fund Platform's Architecture](https://www.dipankar.co/articles/mutual-fund-platform-technical-deep-dive-architecture/)
- [NoCarbon's Market Impact and Future](https://www.dipankar.co/articles/nocarbon-market-impact-sustainable-future/)
- [A Technology Platform for Mutual Funds](https://www.dipankar.co/articles/revolutionizing-mutual-fund-management-comprehensive-technology-platform/)
- [NoCarbon's Waste-to-Energy Innovation](https://www.dipankar.co/articles/nocarbon-technical-innovation-waste-energy-solutions/)
- [NoCarbon: Waste Management and Clean Energy](https://www.dipankar.co/articles/nocarbon-revolutionizing-waste-management-clean-energy-india/)
- [Data-Driven Route Optimization at Blackbuck](https://www.dipankar.co/articles/data-driven-route-optimization-blackbuck-trucking-revolution/)
- [Scaling Infrastructure for Real Estate Tech](https://www.dipankar.co/articles/scaling-real-estate-tech-optimizing-infrastructure-high-growth/)
- [Zambia's CBD Project: Socioeconomic Impact](https://www.dipankar.co/articles/zambia-cbd-project-socioeconomic-impact/)
- [Optimizing Database Performance at Proptiger](https://www.dipankar.co/articles/scaling-success-optimizing-database-performance-proptiger/)
- [The Technical Side of CBD in Zambia](https://www.dipankar.co/articles/zambia-cbd-cultivation-technical-aspects/)
- [A Real-Time SEO Rank Tracking Platform](https://www.dipankar.co/articles/revolutionizing-seo-analytics-scalable-real-time-rank-tracking-platform/)
- [Zambia's CBD Cultivation Opportunity](https://www.dipankar.co/articles/zambia-cbd-cultivation-economic-opportunity/)
- [Data Insights at Momspresso](https://www.dipankar.co/articles/data-insights-transforming-momspresso-content-strategy/)
- [A Loyalty Points Management System](https://www.dipankar.co/articles/innovating-customer-engagement-loyalty-points-management-system/)
- [Momspresso's New Recommendation Engine](https://www.dipankar.co/articles/powering-personalized-content-momspresso-new-recommendation-engine/)
- [Quiki's Progress in Zambia](https://www.dipankar.co/articles/quiki-driving-social-economic-progress-zambia/)
- [Quiki's Franchise Model for Entrepreneurs](https://www.dipankar.co/articles/quiki-franchise-model-empowering-local-entrepreneurs-cutting-edge-technology/)
- [Modernizing a Marketing Automation API](https://www.dipankar.co/articles/modernizing-marketing-automation-api-redesign-multi-language-integration/)
- [Quiki's Ride-Matching Algorithm](https://www.dipankar.co/articles/quiki-advanced-ride-matching-algorithm/)
- [The Technology Behind Quiki in Zambia](https://www.dipankar.co/articles/quiki-technology-powering-zambias-mobility-revolution/)
- [Scalable Backend Services for Set-Top Boxes](https://www.dipankar.co/articles/developing-scalable-backend-services-next-generation-set-top-boxes/)
- [A Scalable Data Pipeline for Momspresso](https://www.dipankar.co/articles/building-scalable-data-pipeline-momspresso-empowering-content-personalization/)
- [Quiki: A Ride-Sharing Platform for Cities](https://www.dipankar.co/articles/quiki-innovative-ride-sharing-platform-revolutionizing-urban-mobility/)
- [Quiki's Smart Transportation for Zambia](https://www.dipankar.co/articles/quiki-revolutionizing-mobility-zambia-smart-transportation-solutions/)
- [A Recommendation System for Lenskart](https://www.dipankar.co/articles/revolutionizing-ecommerce-recommendation-system-lenskart-eyewear/)
- [Building a Widget Platform for 99Acres](https://www.dipankar.co/articles/accelerating-frontend-development-widget-platform-99acres/)
- [An E-commerce Platform with Custom Payments](https://www.dipankar.co/articles/building-scalable-ecommerce-platform-custom-payment-integration/)
- [Digital Infrastructure for Indian Websites](https://www.dipankar.co/articles/revolutionizing-digital-infrastructure-leading-indian-websites/)
- [Ubermens: An IQ Quiz and Reward Platform](https://www.dipankar.co/articles/gamifying-intelligence-ubermens-iq-quiz-reward-platform/)
- [An Integrated ATS Widget at thehiringtool](https://www.dipankar.co/articles/revolutionizing-recruitment-integrated-ats-widget-thehiringtool/)

## Contact
Email: contact@dipankar.co
LinkedIn: https://www.linkedin.com/in/dipankarsarkar
ORCID: https://orcid.org/0000-0001-5431-6367