AI/ML and full-stack engineering, built to ship.

I design and build production systems where models, data and the web layer meet. Based in Ahmedabad, working with teams in Gandhinagar, Bangalore and remotely.

Open to full-time AI/ML and full-stack roles, plus focused contract engagements.

Engineering services

AI/ML Engineering

Models that leave the notebook and survive production traffic, retraining and messy real-world data.

  • Dataset design, preprocessing and evaluation pipelines
  • Model training, fine-tuning and error analysis
  • Inference services with monitoring and rollback paths

Python · PyTorch · TensorFlow · scikit-learn · FastAPI

Computer Vision

Vision systems for classification, detection and inspection, from MRI scans to agricultural imagery.

  • Vision Transformer and CNN pipelines
  • Preprocessing, augmentation and annotation strategy
  • Accuracy, latency and failure-mode reporting

OpenCV · Vision Transformers · PyTorch · Gradio · NumPy

LLM & Agent Systems

Retrieval, tool use and evaluation for products that need reliable language-model behaviour, not demos.

  • RAG pipelines with grounded citations
  • Agent tooling, guardrails and evaluation harnesses
  • Latency, cost and context-window optimisation

LLM APIs · RAG · Vector search · Python · Node.js

Full-Stack Product Builds

End-to-end product delivery: the API, the database, the interface and the deployment path.

  • Django, Node and React application builds
  • Relational schema and API design
  • Performance, accessibility and release hardening

React · Next.js · Django · PostgreSQL · Redis · Docker

Locations

  • Ahmedabad — Home base. Available for on-site or hybrid work across the city.
  • Gandhinagar — Regularly available for on-site delivery, reviews and workshops.
  • Bangalore — Available for project travel and remote collaboration with product teams.
  • Remote — Async-friendly across IST, with overlap for EU and US-East teams.

How an engagement runs

  1. Brief — A 30-minute call to understand the problem, constraints and what success looks like.
  2. Scope — A short written plan: architecture, milestones, risks and what is explicitly out of scope.
  3. Build — Weekly working sessions and visible progress, with the codebase and decisions documented as we go.
  4. Handover — Tests, deployment notes and a walkthrough so the system can be maintained without me.

Ways to work together

  • Full-time role — Joining an in-house AI/ML or product engineering team in Ahmedabad, Gandhinagar, Bangalore or remote.
  • Project sprint — A scoped 2–6 week build with a clear deliverable, milestone plan and handover documentation.
  • Technical advisory — Architecture reviews, model evaluation and delivery support for an existing team.

Questions

Are you available for full-time roles?

Yes. I am open to full-time AI/ML and full-stack engineering roles, as well as focused contract engagements. The fastest way to start is a short email with the problem, the team and the timeline.

Which locations do you work in?

I am based in Ahmedabad and regularly work with teams in Gandhinagar and Bangalore. I also take remote engagements across India and with teams in Europe and the United States, with overlap scheduled around IST.

What kinds of projects do you take?

Computer vision systems, LLM and retrieval pipelines, and full-stack product builds where the model and the product need to work together. I am most useful when there is a real deployment constraint: latency, cost, reliability or messy data.

Can you join an existing codebase?

Yes. I start with a code and architecture review, identify the highest-risk gaps, and work inside the team's existing conventions rather than rewriting for its own sake.

How do engagements usually start?

A 30-minute call, followed by a written scope with milestones, assumptions and a clear definition of done. No long discovery phase before you see a plan.

Do you offer fixed-price work?

For well-defined deliverables, yes: a milestone-based fixed scope. For exploratory model work or evolving product requirements, a weekly or monthly engagement is usually fairer to both sides.