Adarsh Bhardwaj

AI Engineer — India

Building AI for the two places a wrong answer costs you: money and health.

I'm founding Paisa, an AI CFO and financial operating system, and I built NutriScan AI, a health agent with a clinical safety layer under it. Before that I worked on MCP-based agent tooling at MCPfy.ai. I care about the boring part — making the model's answer correct, then making it fast.

What's behind this page. The shape Paisa and NutriScan both have, running live in 3D. A question arrives at the top, a supervisor hands it to whichever specialist owns it, and the work drops into the engine below — where the lines are straight because that layer is deterministic and the model has no vote in it. Move the pointer, or scroll.

Selected work

Four systems, one rule: the model explains, it does not decide the number.

Paisa

July 2026 — present

AI CFO & financial operating system

Most finance assistants will happily invent a number. Paisa can't — the language model never touches arithmetic. A tested engine computes every figure (expenses, UPI, SIPs, loans, investments, tax) and the model's only job is to explain what the engine returned. Above that sit four agents with separate remits — accounting, tax, investing, planning — and a router that decides which one should answer. Bank accounts connect through secure connectors, so the numbers being explained are real ones.

Built withPython · FastAPI · React Native · TypeScript · AWS · Docker · Kubernetes

Repository

NutriScan AI

2026

AI health agent — clinical reasoning, in production

An AI health agent that reasons over symptoms, lab results and clinical history. Five specialists — doctor, labs, fitness, nutrition and coach — sit behind a supervisor that routes each question to whichever one should own it, and each specialist only sees the slice of the patient record it needs. Underneath the chat is the part that matters: a deterministic triage engine of forty clinical rules that runs before the model reasons, fails closed on error, and can end the turn outright. A model may raise suspicion; nothing can lower a rule-derived verdict. Insights carry their own certainty, derived from sample size rather than asserted, and any number the agent quotes is computed in tested code rather than produced by the model. 156 tests.

Built withTypeScript · Next.js · Anthropic & OpenAI APIs · Python · FastAPI · PyTorch · PostgreSQL · React Native · Docker

nutritiscan.com Repository

Agent framework

2025

Multi-agent orchestration, pulled out of Paisa

The scaffolding underneath Paisa's agents, extracted to stand on its own: role-scoped agents, tool calling, an intent router that maps a question to the agent that should own it, and outputs strict enough that the rest of the system can depend on them instead of re-parsing prose.

Built withPython · LangChain · tool calling · JSON schema

RAG financial assistant

2025

Question answering over statements and filings

Ingestion and chunking tuned for tables rather than paragraphs, embeddings into a vector store, and answers that cite the row they came from. The citation is the whole point — an unsourced number is worse than no answer.

Built withPython · embeddings · vector search · LangChain · PostgreSQL

Experience

Founder & AI Engineer, Paisa

July 2026 —

Building the product and the company. Agent architecture, secure bank integrations, the financial modules, the backend and the deploy pipeline — and presenting the whole thing to mentors and investors.

Python · FastAPI · JavaScript · AWS · Docker · Kubernetes

AI Engineer, MCPfy.ai

Internship · Aug 2026

MCP-based developer tooling. Built and integrated the APIs that agent workflows ran on, chased down reliability bugs, and tested the integrations other people's agents depended on.

Python · TypeScript · MCP · REST APIs · Git

AI Engineer, NutriScan AI

2026

Built the AI health agent end to end: the specialist roster and the supervisor that routes between them, and the clinical safety layer underneath — deterministic triage, structured clinical state, fail-closed escalation. Food-image and nutrition analysis is one of the signals it reasons over, not the product.

TypeScript · Next.js · Anthropic API · Python · FastAPI · PyTorch · PostgreSQL · React

Open source contributor

Ongoing

Features, bug fixes, documentation and performance work across AI/ML and developer tooling — mostly Python, JavaScript and agent frameworks.

GitHub · Hugging Face · LangChain

Stack

Languages
Python, TypeScript, JavaScript, Java, C++, SQL
AI & LLM
LangChain, OpenAI API, Anthropic, MCP, prompt design, tool calling, structured output
Agents & RAG
Multi-agent orchestration, retrieval pipelines, vector search, embeddings, intent routing, text-to-SQL, grounded generation
ML
PyTorch, TensorFlow, scikit-learn, NumPy, Pandas, Streamlit
Backend
FastAPI, Django, Node.js, Express, REST, GraphQL, WebSockets
Frontend
React, Next.js, React Native, Expo
Data
PostgreSQL, MongoDB, Redis, DynamoDB, vector databases
Infra
AWS, GCP, Lambda, Docker, Kubernetes, GitHub Actions, Jenkins, CI/CD
Practice
Git, testing, monitoring, code review, Agile, Jira

Background

Contact

Open to AI engineering roles and contract work.

Email is the fastest way to reach me. Happy to talk about agents or retrieval even when it isn't a pitch.

adarshbhardwaj9182@gmail.com