Open to interesting problems

Niraj Kumar Jha

Full Stack Engineer/Punjab, India

Full Stack Engineer building AI products end to end - pipelines, integrations, and the boring plumbing that makes them survive production.

NJNiraj Kumar Jha - Full Stack Engineer
Years shipping
3+

Years shipping

Products in production
6

Products in production

Users served
15k+

Users served

About

I'm a Full Stack Engineer at Creatr, where I build the AI systems behind DeepBuild - an email pipeline wired into Gmail webhooks, M365 and MSP integrations, and a retrieval-backed pipeline for long-form reports. Before that I was the founding engineer at EventEye, taking an event management and discovery platform from an empty repo to 15,000+ users.

On the side I ship products of my own, most recently The Coder'z, an AI learning platform built around a retrieval system that turns a developer's own work into a personal agent. I care about designing the data model before writing the code, keeping AI behind guardrails instead of hoping for the best, and shipping things people actually use.

Experience

  1. Creatr

    Current
    Full Stack Engineer·April 2026 - Present·Remote
    • Built the AI email pipeline on Gmail webhooks: form intake routes to the right sender identity, gathers evidence, and has Claude draft an on-brand email in ~1 minute for admin approval - with pixel-tracked opens.
    • Built DeepBuild core modules and reusable integration packages - Gmail webhooks, email tracking, full M365 (auth, email, calendar), Atera and Freshdesk.
    • Owned MSP integrations and onboarding end to end, including M365 email and license provisioning plus a self-healing agent that auto-handles software tasks and falls back to human review when unsafe.
    • Shipped a long-form report pipeline that embeds uploaded documents so Claude retrieves sections on demand while writing, plus a guarded product assistant with multi-layer checks on every write action.
  2. EventEye

    Founding Engineer
    Founding Engineer·June 2025 - Feb 2026·Remote
    • Built and scaled a full-stack event management and discovery platform from scratch, serving 15,000+ users.
    • Designed the backend for listings, ticketing, and team-based registrations, plus admin dashboards, real-time workflows, secure payment flows, and queue-backed certification pipelines.
    • Worked with a team of 3-4 engineers on feature planning, code review, and system design decisions.
  3. Growth Engineer Intern·Jan 2025 - May 2025·Remote
    • Built and deployed 4-5 customer-facing AI products, including podcast generation and poster creation.
    • Developed LLM-powered workflows for personalized content delivery in an e-learning platform, and multi-step AI pipelines integrating external tools and APIs.
    • Improved reliability through API optimization and debugging, and implemented secure authentication for a text-to-code platform.

Selected Work

  • The Coder'z - AI-powered ed-tech platform for developers screenshot

    The Coder'z

    Jun 2025 - Jan 2026

    AI-powered ed-tech platform for developers

    A learning platform with structured pathways, project generation, and portfolio tooling, built around a RAG system called 'KnowMe' that turns a developer's own work into a personal AI agent.

    • End-to-end pipeline: data → embeddings → retrieval → LLM response generation.
    • Tuned context retrieval to reduce hallucination and improve answer accuracy.
    • Next.js
    • TypeScript
    • PostgreSQL
    • Prisma
    • Embeddings
    • Vector Search
  • ValidateX - Idea validation with on-chain rewards screenshot

    Idea validation with on-chain rewards

    A community-driven platform where people vote on startup ideas and earn tokens for feedback that actually holds up.

    • Next.js
    • TypeScript
    • Web3
    • PostgreSQL
    • Prisma
  • AfterClass - Campus-first social and study rooms screenshot

    Campus-first social and study rooms

    Float between rooms - your people are already there. Real-time study rooms, live hangouts, and college community features.

    • Next.js
    • TypeScript
    • WebSockets
    • PostgreSQL
    • NextAuth
  • Vani - Learning from long-form audio screenshot

    Vani

    2025

    Learning from long-form audio

    An MVP for processing large audio files and turning them into something you can actually search and learn from.

    • Next.js
    • TypeScript
    • Audio Processing
    • AI

Stack

Languages

  • TypeScript
  • JavaScript
  • C++
  • C
  • SQL

Frontend

  • Next.js
  • React.js
  • Remix
  • TailwindCSS
  • Shadcn UI
  • TanStack Query
  • Framer Motion

Backend

  • Node.js
  • Express
  • Server Actions
  • REST APIs
  • WebSockets
  • Background Workers

Data

  • PostgreSQL
  • pgvector
  • MongoDB
  • Redis
  • Prisma

AI

  • Claude
  • OpenAI
  • RAG & Embeddings
  • Agent Pipelines
  • Cursor

Infra & Tooling

  • Docker
  • AWS
  • Vercel
  • Git
  • Postman
  • PostHog

Currently Learning

  • Distributed systems - consensus, replication, and failure modes under real load
  • Retrieval quality - chunking, reranking, and evaluating RAG instead of eyeballing it
  • System design - designing for the second year of a product, not just the first launch

Writing

/8 min/4

The like button that lied to me on purpose

I turned the network off, liked my own post, and watched the count go up anyway. The lie was deliberate and correct here. Where it stops being correct is the whole question.

  • Next.js
  • Performance
Featured
/13 min/9

The same question, four thousand ways

A support bot answered ninety questions four thousand times and paid full price for every one. Exact-match caching hit two percent. Here is what actually works.

  • AI
  • Engineering
/10 min/5

The login that allocated 32 MB

Better Auth hashes passwords with scrypt, which wants 32 MB of memory per login. On Cloudflare Workers that is a problem. Replacing it was the easy part.

  • Security
  • Engineering
/6 min/1

Why your loading.tsx never showed up

The skeleton took an afternoon and nobody ever saw it. The file was in the right folder with the right name, and four separate things stopped it rendering.

  • Next.js
  • Performance
/8 min/4

The like button that lied to me on purpose

I turned the network off, liked my own post, and watched the count go up anyway. The lie was deliberate and correct here. Where it stops being correct is the whole question.

  • Next.js
  • Performance
Featured
/13 min/9

The same question, four thousand ways

A support bot answered ninety questions four thousand times and paid full price for every one. Exact-match caching hit two percent. Here is what actually works.

  • AI
  • Engineering
/10 min/5

The login that allocated 32 MB

Better Auth hashes passwords with scrypt, which wants 32 MB of memory per login. On Cloudflare Workers that is a problem. Replacing it was the easy part.

  • Security
  • Engineering
/6 min/1

Why your loading.tsx never showed up

The skeleton took an afternoon and nobody ever saw it. The file was in the right folder with the right name, and four separate things stopped it rendering.

  • Next.js
  • Performance
/6 min/2

Three layers people call a server action

A two line pull request nearly went through on my phone. It was the third delete button that month, written the third different way, and one of them was wide open.

  • Next.js
  • Interviews
Featured
/7 min/8

The prefetch that made the list page slower

A three line accessibility fix made navigation instant and the orders list slow. The obvious next move made it four times worse, and the reason is one default.

  • Next.js
  • Performance
/6 min/2

Three layers people call a server action

A two line pull request nearly went through on my phone. It was the third delete button that month, written the third different way, and one of them was wide open.

  • Next.js
  • Interviews
Featured
/7 min/8

The prefetch that made the list page slower

A three line accessibility fix made navigation instant and the orders list slow. The obvious next move made it four times worse, and the reason is one default.

  • Next.js
  • Performance

Education

  1. B.Tech, Computer Science·May 2022 - May 2026·7.5 / 10 GPA
  2. Kathmandu Model Higher Secondary School

    Higher Secondary (10+2)·Apr 2020 - May 2021·Science
  • Meta Front-End Developer Certification
  • Master Data Structures & Algorithms in C/C++ - Abdul Bari
  • Winner - Intra-University Public Speaking Competition
  • Languages: English, Hindi, Nepali

Contact

Building something interesting?

I'm always up for talking about AI products, backend design, or a problem that's been stuck for a while. Email is fastest - DMs on X work too.

jhaniraj45@gmail.com