Resume

Principal / Lead AI Engineer | Architect & Fractional CTO | Founder | ex-Amazon, ex-Agoda

⬇ Download Full 2-Page Resume ⬇ Download 1-Page Resume

Contact

📧

Email

Fastest response

📱

Phone

IST 10 AM – 7 PM

💼
💻

Summary

  • AI engineer and principal-level architect who builds production agentic systems end-to-end — LangGraph agent graphs, RAG pipelines, voice agents, and multi-tenant backends — currently live with 5 paying CBSE schools via Yuvan.
  • 10+ years shipping production systems at Amazon, Agoda, and high-growth startups across India, Southeast Asia, and the US; brings senior-engineering rigor to AI — distributed systems, scale, security, cost engineering.
  • Runs LangSmith evals, retrieval-precision metrics, and agent traces before calling anything production-ready — engineering discipline, not vibes-based prompting.
  • Equally comfortable leading teams (15+ engineers, INR 30Cr P&L) and building solo as a senior freelancer / fractional CTO embedding agentic features inside someone else's stack.

Core Skills

Agentic AI and LLMs

LangChain, LangGraph (stateful agent graphs), RAG, semantic FAQ caching, function/tool calling, prompt engineering, evals, agent observability (LangSmith), guardrails and verifier patterns, multi-agent orchestration

LLM Providers and Voice

OpenAI GPT-4o / 4o-mini, Realtime API, Embeddings, Anthropic Claude, Gemini, OpenAI Whisper, WebRTC voice agents, streaming + interruption handling

Vector and Retrieval

pgvector, embeddings (text-embedding-3, Cohere, BGE), hybrid search, chunking strategies, cosine-similarity caching, retrieval evals

AI-Native Backend

Python 3.11+, FastAPI (async), Pydantic v2, asyncpg, pytest-asyncio, Java 21, Spring Boot, Spring Security, Hibernate JPA, Scala

Architecture

Microservices, distributed systems, event-driven (Kafka), HLD/LLD, design patterns, SOLID, clean code, API design (REST + OpenAPI), TDD

Data

PostgreSQL, pgvector, Cassandra, DynamoDB, Redis, MongoDB, SQL tuning, PgBouncer, Supabase (Auth + RLS)

Cloud and DevOps

AWS Certified Solutions Architect — S3, ECS, Lambda, IAM, Serverless | Docker, Kubernetes, GitHub Actions, Jenkins, Railway, Vercel, Grafana, ELK

Frontend

Next.js 15 (App Router), TypeScript, JavaScript, HTML, CSS, React, Tailwind, shadcn/ui, KaTeX

Testing and QA

Test-Driven Development, JUnit, Mockito, pytest, pytest-asyncio, httpx

AI-Assisted Development

Claude Code, Cursor, GitHub Copilot, v0.dev — shipping with AI as a daily collaborator

Leadership

Strategic planning, hiring and team growth, mentoring, curriculum design, stakeholder management, agile/scrum, cross-functional delivery, founder-mode product + GTM

Other

Data Structures and Algorithms, REST, system design

Experience

Founder and AI Engineer | Yuvan (Voice-First AI Math Tutor)

Agentic AI | LangGraph | RAG | OpenAI Realtime | FastAPI | Next.js | Supabase | pgvector

– Present | Remote
  • Built and shipped Yuvan, a voice-first AI tutor for CBSE Class 10 students — onboarded 5 paying schools as design partners on a B2B2C model (school recommends, parent pays); product currently live in MVP with paid pilots.
  • Designed the full agentic architecture: a LangGraph state machine per doubt — embed → semantic FAQ-cache lookup → RAG retrieval over NCERT chunks → GPT-4o-mini generation → math verifier → topic auto-tagger → persistence — with an explicit re-explain loop on detected student confusion.
  • Engineered a WebRTC voice pipeline on OpenAI Realtime (STT + TTS + VAD + barge-in interruption) with backend-minted ephemeral tokens; achieved sub-1.5s p95 first-audio latency in pilot sessions while keeping all reasoning, RAG, and verification server-side.
  • Built a RAG layer on Postgres + pgvector (Supabase) — ingestion, chunking, OpenAI embeddings, top-k retrieval, citation-grounded answers — and a global semantic FAQ cache (cosine >= 0.92) targeting ~30% hit rate to push per-session OpenAI cost below the unit-economics ceiling.
  • Implemented a math-verifier safety net that re-checks every numerical claim before the AI speaks it (under 1s overhead) — the difference between a demo and a tutor that schools trust.
  • Built the full multi-tenant product: 4 user roles (Student / Parent / School Admin / Super Admin), Supabase Auth + Postgres RLS for tenant isolation, parental-consent flow (DPDP / NCPCR-aligned), weekly parent reports, and a school engagement dashboard.
  • Owned everything end-to-end: architecture, code, infra (Railway + Vercel + Supabase under $10/mo demo budget), LangSmith tracing, evals, school sales calls, parent-onboarding flows, pricing, and pilot success.

Director, Full-Stack Engineering | PhysicsWallah

Java | Spring | Microservices | Python services | AWS | AI dev tooling

– | Bengaluru / Indore
  • Hired and led a team of 15 engineers to build an in-house Learning Management System using Java, Spring, Hibernate, TypeScript, microservices, and AWS.
  • Stood up CI/CD on Jenkins + Docker, lifting deployment frequency by 15% and cutting release-night fire-drills.
  • Introduced Python/FastAPI services for ML-adjacent workloads (analytics + content tagging) alongside the core Java stack, and rolled out AI-assisted development (Copilot, code review automation) across the team.

Independent AI Engineer and Builder | Agentic AI R&D

LangChain multi-agent patterns | voice pipeline prototyping | embeddings benchmarking

– | Remote
  • Pivoted full-time into AI engineering and shipped an early AI math-teacher prototype to 900+ users — the user-validation work that proved demand and became Yuvan.
  • Benchmarked LangChain multi-agent patterns (router, supervisor, plan-and-execute) on real student conversations to pick the architecture Yuvan now ships on.
  • Prototyped four voice-agent stacks (OpenAI Realtime, Whisper + ElevenLabs, Whisper + Coqui, Gemini Live) and ran latency/cost shootouts — the data behind picking Realtime + WebRTC.
  • Cost-engineered embeddings choice across OpenAI text-embedding-3, Cohere v3, BGE for retrieval precision vs. cost; documented trade-offs that drove the FAQ-cache design.
  • Side build: production-grade microservices flash-sale system (Java 20, React, TypeScript) to keep distributed-systems skills sharp under high-concurrency load.

Senior Engineering Manager (Contract) | ForaySoft (client: Wayfair)

Java | Spring Security | Hibernate | TypeScript | PostgreSQL | Kubernetes | AWS

– | Remote
  • Led 10 engineers on a fixed-term migration of Wayfair's seller-promotions microservice — Java, Spring Security, Hibernate, PostgreSQL, Docker, AWS — cutting system failures by 28%.
  • Re-architected for scale on Kubernetes, supporting a 25% increase in transaction volume during peak e-commerce promotions.

Vice President, Engineering (Hands-on) | Nolan EduTech

Java | Spring | JPA | Microservices | AWS | TypeScript | Curriculum + GTM

– | Remote
  • Launched a coding-bootcamp vertical from zero to INR 30 crore (~$3.6M) in revenue inside one year — owned the curriculum, the platform, and the live teaching.
  • Designed end-to-end backend + full-stack programs aligned to industry hiring; hit a ~95% placement rate across cohorts in the thousands.
  • Re-architected an internal HackerRank-style assessment platform on Java + Spring Boot + AWS for scale and reliability under load.

Lead Developer | CoinSwitch Kuber

Python | FastAPI | Django | PostgreSQL | Redis | AWS | Microservices

– | Remote
  • Migrated a monolithic crypto-trading app to microservices on Python, FastAPI, and Django on AWS — the same async Python stack used for AI services today.
  • Rolled out Redis query caching to drop query latency by 30%, and tuned PostgreSQL with PgBouncer connection pooling, cutting query latency a further 40% and unblocking connection-saturation incidents.
  • Held the system stable through a user surge from 600K to 10M during a crypto bull run.

Software Development Engineer II | Amazon

Java | DynamoDB | AWS | Payments / Digital Goods

– | Chennai
  • Hardened payment security against CSRF attacks, reducing impact by 19.7%.
  • Architected a throttling system on DynamoDB + Java that cut coupon-code misuse by 15.3% in the first month post-launch.
  • Integrated the Amazon Games payment flow into the Digital Payments Platform; improved payment-page accessibility for differently-abled users and search engines across geographies and devices.

Contract SDE | Porch.com, Domino, others

Python | Node | MongoDB | Facebook Marketplace API

– | Seattle / Bengaluru
  • Built a service from scratch integrating Porch.com's backend with Facebook Marketplace — doubled service-quote requests and lifted revenue 13% by tapping ~200M FB users.
  • Optimized MongoDB analytics sorting via Python tooling for Domino, improving response time 30%.

Senior Software Developer | Agoda.com

Scala | Cassandra | Kafka | Distributed systems

– | Bangkok
  • Engineered an in-house A/B testing pipeline processing millions of messages on Scala + Cassandra + Kafka — cut time-to-insight for product managers by 22%.
  • Wrote shared Scala libraries that grew tracked-user coverage by 10%.

Software Engineer | Hadoolytics Technologies

Java | Spring | AWS | Cassandra

– | Gurugram
  • Architected a real-estate platform from scratch on Java, Spring, AWS, and Cassandra.
  • Integrated Swagger for auto-generated API docs, saving ~20% of development time per cycle.

Education

M.Tech, Computer Science and Engineering

NIT Jalandhar | India — CGPA 7.5/10

  • Taught and mentored thousands of undergraduate students as a teaching assistant / instructor — the first chapter of a teaching thread that later became Nolan EduTech and PhysicsWallah.

B.E., Computer Science and Engineering

Certifications

  • AWS Certified Solutions Architect

Selected Project Highlights

  • Yuvan — voice-first AI tutor; LangGraph agent | RAG on pgvector | OpenAI Realtime WebRTC | math verifier | semantic FAQ cache | multi-tenant Supabase RLS | 5 paying schools. Architecture deep-dive and code samples available on request.
  • Flash-sale system — production-grade microservices simulating real concurrency, queueing, and back-pressure on Java 20 + React + TypeScript.
  • Online assessment platform — HackerRank-style code-execution and grading on Java + Spring Boot + AWS, scaled for thousands of concurrent test-takers.