Hj HIMANSHU JAIN
Chemistry Faculty · AI Product Builder · Teacher who ships

A teacher who
ships products.

A decade teaching JEE Chemistry. These days I build the AI tools I always wished my students had — and I love getting them into people's hands. 15+ deployed products so far. This page shares what I've built, and a little of how I think while building it.

See how I think → The products
15+
Live products
10+
Yrs teaching
JEE
+ CBSE domain
AI
native stack
∞Hj
Himanshu Jain
Specimen · HJ-01Teacher → Builder
Reactant
Explaining a
reaction mechanism
// break a hard idea into ordered, testable steps
Product
Writing a PRD
// break a hard problem into ordered, testable steps

The classroom was my first product team. Ten years at Aakash, FIITJEE, and Amazon Academy — plus two institutes I built from scratch — taught me to notice the exact moment a learner gets stuck, and to find the smallest thing that helps them through it.

That habit turned out to be product management. So I started building the tools I wished my students had: RAG doubt-solvers, answer-grading pipelines, voice trainers, GTM agents — on a real stack, used by real students. I enjoy taking an idea all the way to something people can actually use.

Today I'm finishing HelloPM's AI Product Management program. The teacher-to-builder path has been a natural one for me — and I'm grateful for where it's led.

01 /

Reaction Mechanism

How one product actually got built

Most portfolios show what someone built. I'd also like to share how I think while building. So here's ChemIQ — the product I've put the most into — laid out the way I'd walk a student through a reaction on the board: one step leading into the next, including the step that didn't work at first.

1 Reactant

The problem I lived

Students stuck on Chemistry at 11pm. No teacher awake, generic chatbots hallucinate, textbooks don't answer this doubt. I watched it for a decade.

2 Insight

Why generic AI fails here

JEE answers must cite the syllabus, not the internet. The doubt is often a photo of messy handwriting on a dark slide. Retrieval + vision, not a wrapper.

3 Build

The decision that mattered

Hybrid BM25 + vector retrieval over my own notes, a Groq vision model to read handwritten slides, and a multi-stage grading pipeline — answers are sourced, not guessed.

4 Side-reaction

What broke

Pure vector search returned plausible-but-wrong passages on lookalike reactions. Latency spiked on long PDFs. Both surfaced only once real students used it.

5 Product

What I learned

Evals aren't a finishing step — they're the product. RAG failure analysis now lives in the PRD before I write a line. Domain depth is the moat, not the model.

02 /

The Periodic Table of Builds

Position = strategy

Not just a grid. Columns are clusters (where I build), rows are maturity (shipped → in progress → concept). You can read my whole product strategy in one glance.

Edtech & Learning Career & Tools PM Builds ↓ rows = maturity
03 /

Field Notes

Live from Substack · "Chemistry with HJ"
Loading latest notes…
04 /

HelloPM Coursework

Assignments & artifacts from the program
HelloPM AI Product Management Program completion certificate
Certificate of Completion

AI Product Management Program

A four-month live program covering AI product strategy, PRDs, RAG systems, agentic tools, and end-to-end product thinking.

Issued June 2026 · HelloPM · ID 2H3Q8L
View certificate ↗
Assignment 1 · Product Teardown

redBus — Flows, Frictions & Reverse-Engineering

A full teardown of redBus: mapping every flow and friction point, reverse-engineering key features, segmenting users, and prioritising problem hypotheses — backed by real screenshots and wireframes.

Product AnalysisUser SegmentationFeature Reverse-EngineeringPrioritisation
Read the teardown → Prototype ↗ Phase 1 Phase 2 Phase 3 Wireframes Deck
Assignment 2 · PRD

Product Requirements Document

A complete PRD written to program spec — problem framing, goals, user stories, scope, and success metrics for a defined product build.

PRDProduct SpecSuccess Metrics
Read PRD ↗
Assignment 3 · Technical PRD

ChemIQ — RAG System PRD

The product requirements for ChemIQ's retrieval-augmented generation system: grounding a JEE-Chemistry assistant in real lecture notes, with retrieval design and evaluation baked in.

RAGAI ProductTechnical PRDEvaluation
Read the case study → Read PRD ↗
Assignment · AI Adoption

RoleSpec — From AI Curiosity to AI Usage

An exploration of how professionals move from experimenting with AI to depending on it — and what that shift means for building tools people actually adopt.

AI AdoptionProduct ThinkingGo-to-Market
Read ↗
Assignment · Agentic Product

Vela — AI Day Planner

An agentic day-planning concept built for the Indian context: turning intent into a structured, realistic daily plan. Presented as a full product deck.

Agentic AIProduct DesignConcept
View deck ↗ Live ↗
05 /

Ask My Portfolio

Answers from my real work

Curious about my work? Just ask.

A recruiter, a founder, a fellow builder — ask anything about what I've built, my stack, or the teacher-to-PM story. It answers from my actual portfolio.

What's your strongest AI product? How does teaching make you a better PM? Tell me about ChemIQ. What's your tech stack?
Demo mode — answers from a built-in summary. Add a free Groq key on deploy to make it conversational.