Full-stack engineerAI agents · async pipelines · consumer products

Shrijit.

Full-stack engineer with 3 years in production. I build AI agents, async pipelines and consumer products, take them from zero to one, and own them end to end.

Now → Full Stack Software Developer at M32 Labs, San Francisco (remote)

See the work shrijit.m00@gmail.com

3 yrsin production

3live AI products

0 → 1ship it, own it

AI agents ✦ Voice ✦ RAG pipelines ✦ Async workers ✦ Multi-tenant SaaS ✦ Web3 ✦ Zero to one ✦AI agents ✦ Voice ✦ RAG pipelines ✦ Async workers ✦ Multi-tenant SaaS ✦ Web3 ✦ Zero to one ✦AI agents ✦ Voice ✦ RAG pipelines ✦ Async workers ✦ Multi-tenant SaaS ✦ Web3 ✦ Zero to one ✦

TypeScript ✦ Python ✦ LangGraph ✦ Next.js ✦ NestJS ✦ PostgreSQL ✦ Redis ✦ Kafka ✦ Kubernetes ✦TypeScript ✦ Python ✦ LangGraph ✦ Next.js ✦ NestJS ✦ PostgreSQL ✦ Redis ✦ Kafka ✦ Kubernetes ✦TypeScript ✦ Python ✦ LangGraph ✦ Next.js ✦ NestJS ✦ PostgreSQL ✦ Redis ✦ Kafka ✦ Kubernetes ✦

01 / RECEIPTS

Numbers from production, not slides.

Measured on live systems: search, inference cost, reliability and latency.

0×

faster search across 368k documents: 9,398 ms down to 154 ms by moving sort into Atlas Search.

~0%

lower AI inference spend after a model swap: $37/day down to $4–7/day.

~0%

scoring reliability, while draining a 20-hour backlog at ~150 per minute.

hallucinations in production across voice agents and RAG pipelines.

<200ms

API p50 with Celery and Redis async workers behind multi-tenant Postgres.

02 / SELECTED WORK

Products I took from zero to live.

Three are live with real users today. Move your cursor over a card to tip it toward you.

● Livesamantha.flashpact.in

Samantha

AI voice agent

A production voice agent: LangGraph multi-agent orchestration, RAG on verified domain data, real-time speech-to-text and text-to-speech over WebSocket, RBAC, audit logging and a red-team eval pipeline. Zero hallucinations. Built and deployed solo.

● Liveinfluensa.xyz

Influensa

AI content intelligence

An AI cofounder for solo content creators: a 4-agent LangGraph system for viral trend discovery, scripts in the creator's own voice, per-scene editing and voice coaching. RAG over pgvector. Open source.

● Live · paying customerscrocko.xyz

Crocko

AI review SaaS for local businesses

For cafés, restaurants and salons: scan a QR code, speak or type in any language, and AI drafts a Google review. Sentiment analytics, WhatsApp notifications and a multi-location dashboard, at ₹299–599/month.

Web + mobile

Swypee

Pet SaaS platform

A consumer pet marketplace with a React web app and a React Native mobile app, vector similarity search under 500 ms, preference matching, RBAC and zero-downtime CI/CD on GCP.

Dashboard + smart contracts

Money Movement

Web3 payments dashboard

A Stripe-style crypto payments dashboard: real-time transaction status over WebSocket, multi-wallet support, fee estimation and Solidity smart contracts for on-chain transfers.

03 / EXPERIENCE

Where I've owned things end to end.

Keep scrolling: each card stacks on the last.

Jul 2026 — Present

M32 Labs

Central AI LLC · San Francisco, USA (remote)

Full Stack Software Developer

Agentic AI for small and medium businesses.

Closed a privilege-escalation hole that exposed every VA screenshot to any logged-in user: a new auth layer across 2 databases, per-VA scoping, 20 tests, verified on live prod with Playwright.
61× search speedup on 368k documents (9,398 ms → 154 ms) by moving sort native into Atlas Search.
Root-caused a fleet-wide outage (a wedged connection pool on the wrong pooler port), then built self-healing auto-recovery, honest /health endpoints and per-service pool budgets.
Cut AI inference spend ~85% ($37/day → $4–7/day) with a model swap, lifted scoring reliability to ~99%, and drained a 20-hour backlog at ~150/min.
Rebuilt a legacy PHP platform from scratch: Node/Express/TypeScript backend, React/Vite/Tailwind/shadcn frontend, JWT, RBAC, 35 tests.

Oct 2024 — Jun 2026

C4i4 Lab

Pune, India

Software Engineer

Built LangGraph multi-agent AI systems for enterprise customers: voice agents with real-time STT/TTS, RAG pipelines and red-team evals, with zero hallucinations in production.
Designed multi-tenant PostgreSQL schemas, Celery/Redis async workers (API p50 under 200 ms), Kafka event pipelines, and AWS/GCP deployments on Docker and Kubernetes.

2019 — 2025

Education

Robotics first, then software that thinks.

2022 — 2025 · Pune

B.Tech, Automation and Robotics

JSPM's Rajarshi Shahu College of Engineering

2019 — 2022 · Amravati

Diploma, Computer Science

Dr. Panjabrao Deshmukh Polytechnic

04 / TOOLBOX

The keys I press every day.

Backend, agents, frontend and infra, plus a little Web3.

Languages

TypeScript
JavaScript
Python
Java
SQL

AI and agents

LangGraph
LangChain
RAG
Multi-agent orchestration
Tool use
Anthropic Claude
OpenAI
Gemini
Prompt engineering
Eval pipelines
LLM-as-judge

Backend

Node.js
NestJS
Express
Fastify
Bun
FastAPI
Flask
REST
GraphQL
WebSocket
Kafka
Celery
SQS
Microservices

Frontend

React
Next.js
Vite
Tailwind CSS
shadcn
TanStack Query
Zustand
Redux
React Native
Framer Motion

Databases

PostgreSQL
MySQL
MongoDB
Redis
Supabase
pgvector
Atlas Search
Drizzle ORM

Cloud and DevOps

AWS
GCP
DigitalOcean
Vercel
Docker
Kubernetes
GitHub Actions
Prometheus
Grafana
Loki

Security

Authorization layers
RBAC
JWT
OAuth
SQL injection prevention
Vulnerability assessment

Web3

ethers.js
Viem
Solana.js
Solidity
MetaMask
WalletConnect

Daily tools

Claude Code
Cursor

05 / CONTACT

Got a zero-to-one problem? Let's build it.

Agents, voice, async pipelines, or a product that needs to exist by next month. Email is fastest.

shrijit.m00@gmail.com +91 95795 83897

GitHub LinkedIn X CreatorPulse · open source