Open to AI/ML, GenAI, and agent engineering rolesKolkata, India

AI engineer · GenAI · agentic systems

Building grounded AI systems people can inspect.

I build grounded AI products and full-stack systems that make complex data useful, inspectable, and reliable. I am looking for a team where careful engineering, evaluation, and useful product delivery matter.

Inspect selected work Download resume

LinkedIn GitHub

System map / working methodLive profile

Principle 01

Evidence first

Ground explanations in inspectable evidence and clear limitations.

Principle 02

Bounded systems

Use deterministic systems around probabilistic model behavior.

Principle 03

Human control

Keep human approval gates around consequential actions.

NM / systems atlas

Traceable decisions over black-box theatre

Current role

AI Engineer Intern

Public proof

FundersAI live

Credential

AWS Cloud Practitioner

Focus

RAG / agents / delivery

Selected work

Projects with status and boundaries visible.

View project archive

01 / Applied AI / Fintech

Deployed

FundersAI

Evidence-first research workspace for Indian stocks and mutual funds.

Built

Built a Next.js and FastAPI product around deterministic comparisons, official AMC-document evidence, retrieval workflows, and a Supabase-first data layer.

Boundary

Research-only; it does not provide personalized investment advice or buy/sell/hold recommendations.

Read case study Open live product

02 / Applied ML / RAG

In progress

CareFlow Intelligence

A synthetic-data, non-clinical research sandbox for bounded RAG, citation validation, and RAFT fine-tuning.

Built

Built bounded document and data agents, citation checks, synthetic FHIR data flows, and a local-model fine-tuning path using public or synthetic sources.

Boundary

Synthetic/public data only; non-clinical; not deployed as a clinical product.

Read case study Read the repository

03 / Agentic Systems

In progress

TalentOS / CareerAgent

A job and career-intelligence system for structured source collection, evaluation, and application preparation.

Built

Designed multi-source collection, deterministic pre-filtering, agent evaluation, application drafting, tracing, and human approval gates.

Boundary

No public deployment is claimed; all application actions remain human-controlled.

Read case study Read the repository

External review context

Hackathon submissions.

FundersAI

OpenAI · Devpost

OpenAI Build Week Hackathon

submittedView Devpost

TalentOS

Google · Devpost

All Things Agentic Hackathon

submittedLink pending

Operating system

How the work stays useful after the demo.

Three working constraints

Evidence first

Ground explanations in inspectable evidence and clear limitations.

Bounded systems

Use deterministic systems around probabilistic model behavior.

Human control

Keep human approval gates around consequential actions.

Career snapshot

A working engineer moving deeper into AI systems.

AI Engineer Intern

Internmo

Aug 2026–Present

Three-month remote AI engineering internship. Currently building Vantara, an end-to-end customer intelligence platform for churn and customer-value prediction, due November 2026.

Building churn-risk and 90-day customer-value models over historical transaction data, using time-based splits to keep evaluation leakage-safe.
Designing customer segmentation that prioritizes high-value at-risk accounts and returns explainable retention recommendations.
Shipping the system as a dashboard and REST API, containerized with Docker, with monitoring, tests, and documentation.

View full timeline

Next conversation

Looking for an AI/ML team where careful systems work has room to compound.

For a role, collaboration, or a technical conversation, the fastest path is email or LinkedIn. The freelance path can come later; the immediate goal is the right full-time engineering team.

Email Naman Message on LinkedIn