Services
Four ways to work together
Every engagement starts with a call and a written proposal: scope, milestones, timeline and a fixed price for each milestone. Prices depend on scope, so I quote each brief rather than list a number that fits nobody.
AI agent sprint
A working LLM agent inside your product, taking real actions safely.
Who it's for
Product teams that want an assistant, a copilot or an automation that does more than chat: it books, files, updates and answers from your own data.
You walk away with
An agent in production, the eval suite that proves it works, and the code in your repository.
What's included
- Tools and integrations over your APIs or MCP, with confirmation gates on anything that writes
- Retrieval (RAG) over your documents and data, and memory where it helps
- Server-side guardrails for the mistakes models actually make: wrong dates, invented facts, duplicate actions
- Evals you can rerun, tracing, and per-conversation cost metering
- Deployed in your accounts, with a handover walkthrough
Product build
From an idea or a prototype to a production v1.
Who it's for
Founders and teams who need a whole product built, not a slice: frontend, backend, data, auth, payments and deployment.
You walk away with
A product your users can sign up for and pay for, running in your accounts.
What's included
- A full-stack web app in TypeScript and React, on Cloudflare or Node with Postgres, or on your existing stack
- Sign-in, roles, payments and billing (Stripe), and an admin view
- CI with typecheck, lint and tests on every change, and preview deploys
- A live preview URL from week one, and a demo every week
- Docs a new engineer can start from
Performance and reliability rescue
Slow queries, timeouts and crashes found, fixed and measured.
Who it's for
Teams whose app has started timing out, running out of memory, double-booking or getting slower every month, and who need it fixed without a rewrite.
You walk away with
A faster, steadier system and a written account of what was wrong and why.
What's included
- Profiling and query plans on your real workload
- Fixes for slow queries, missing indexes, timeouts, memory blow-ups, race conditions and duplicate writes
- A before-and-after report with the numbers
- Checks that stop the same regression coming back
Fractional senior engineer
A senior engineer on your team, part-time, month to month.
Who it's for
Teams that need senior judgement and delivery each week without a full-time hire: features, reviews, architecture and the AI parts nobody else has built before.
You walk away with
Steady senior throughput, and a codebase in better shape each month.
What's included
- Features shipped end to end inside your process
- Code review, architecture and technical decisions written down
- Owning the AI features in production: prompts, evals, cost and incidents
- Mentoring the engineers around me
Specialties
Unusual problems welcome
Beyond the packages, I've shipped work in areas most teams rarely staff for. If your project needs one, say so in the brief.
- Machine learning in the browser on-device models in a Web Worker, no uploads
- Maps and GIS grids, geocoding, deck.gl and MapLibre
- Real-time live rooms on WebSockets and Durable Objects
- 3D on the web three.js scenes and code-generated models
Questions
Before you ask
How do you price?
After a short call I send a fixed price for each milestone, so you know the cost before work starts. Fractional work is a monthly fee for an agreed number of hours.
Who owns the code?
You do. The code lives in your repository and runs in your accounts from the start, and the IP is yours once each milestone is paid.
Will you sign an NDA?
Yes, before we discuss anything confidential.
Where are you, and when can we talk?
Chennai, India, on IST (UTC+5:30). My day overlaps US mornings, UK and European afternoons, and all of Asia and the Middle East.
You build with AI coding agents. Is the code any good?
I run coding agents in parallel, but nothing ships on their word: every change has to pass typecheck, tests and a check in a real browser. That's how one engineer ships at the pace of a small team. You get the tests with the code.
Can you work in our stack?
Most of my work is TypeScript, React and Node, with Python for machine learning, on Cloudflare, AWS, Azure or GCP. If your stack is different, say so in the brief and I'll tell you honestly whether I'm the right person.
How do payments work?
Invoices per milestone, in USD or INR, by bank transfer. For monthly work, at the start of each month.
Do you take small jobs?
Yes, when they're well defined: a stubborn bug, a slow page, an integration. Those usually fit the rescue package.
Have something to build, fix or speed up?
Send a short brief. I reply within one working day, and you get a written proposal with a fixed price within two working days of our call.