Background

About Sumanta: principal engineer, fractional CTO and AI architect

Experience across AI, data and infrastructure, published research, open-source work, and the hardware I'm experimenting on right now.

How I work Expertise About Case studies Play

On this pageExperience Credentials Now

Why experience matters

What my background means for your project.

Experience is only useful if it changes outcomes. Here's how each part of mine turns into something concrete for you.

1M+ transactions a dayBuilt and scaled high-volume systems

→

You avoid the rewriteYour architecture is designed for the load you're heading toward, not the load you have today.

Peer-reviewed researchPublished in Science Immunology

→

Rigour you can measureAI and analytics get evaluation sets, baselines and honest metrics, not demos and anecdotes.

Healthcare data standardsOMOP, OHDSI, FHIR, clinical vocabularies

→

No learning curve on your dimeRegulated, messy clinical data is familiar ground, so delivery starts in week one.

Builder of an AI platformShipped one end to end

→

Founder-level ownershipI've made the trade-offs you're facing: scope, speed, cost, hiring. Advice comes from having shipped.

MongoDB Community LeaderTeaching data modelling in public

→

A stronger team, not dependenceMentoring and documentation are part of the job. Your engineers level up as we build.

Polyglot across 9 layers80+ public repositories, 10+ languages

→

Unbiased technology choicesNo favourite hammer. Each tool is chosen for your problem, your team and your budget.

Credentials

Research, community and open source.

Work that is public, peer-reviewed or open to inspection.

Publication

Science Immunology

Co-author on peer-reviewed research, applying engineering rigour to scientific data.

Read the paper →

Community

MongoDB Community Leader

Recognised contributor to the MongoDB community. Author of an open MongoDB cookbook and aggregation-pipeline tooling.

See the cookbook →

Open source

80+ public repositories

AI pipelines, data tooling, infrastructure helpers, developer CLIs and full products, all open to review.

Browse GitHub →

langchain-rag-productionProduction-grade RAG patterns with LangChain realtime-event-processorHigh-throughput event processing in Python mongodbcookbookRecipes for modelling and querying MongoDB PipelineGeneratorBuild MongoDB aggregation pipelines in Python PyTerraformPython abstraction over Terraform for AWS, Azure, GCP claude-switchGo CLI for AI developer workflows, shipped via Homebrew

Now · AI architect's bench

What I'm building with right now.

Advice on AI is only as good as the hands-on time behind it. These days I'm running large language models on my own hardware, including an NVIDIA DGX Spark, to understand what local, private AI can really do for clients who can't send data to a public API.

currently experimenting

Local LLM inferenceOpen-weight models on DGX Spark: throughput, latency and quality versus hosted APIs.

Private AI for regulated dataOn-prem and air-gapped patterns for finance, healthcare and government-adjacent work.

Fine-tuning & quantisationLoRA adapters and quantised models sized for real hardware budgets.

Agents & MCPTool-using agents, MCP servers and evaluation harnesses for them.

dgx-spark · ~/lab

Open to advisory, fractional CTO & select builds

Let's talk about what you're building.

The fastest route is a free 30-minute call. Prefer to write? Send a message and I'll reply personally within two business days.

Book a 30-min call → Send a message or analyse your challenge first →