AI Product Manager

I start with the teacher in the classroom, then I build.

I sit in the room, watch how people actually use the product, ship with AI tools at a pace that lets me test fast, and cut whatever doesn't help. That is how a simple content player became a teaching assistant used by 100+ schools.

Download resume (PDF) See Learno live Email me
100+schools using Learno
~2 a dayfeatures shipped, tested with teachers, kept only if useful
+95%efficiency on animated videos
+70%efficiency on PPT production

How it started

Dec 2024

A Fusion 360 animation

Think Pro Academy asked me to make 3D models. I only knew Fusion 360, so I animated an engineering assembly of their Lego parts, rendered a video and sent it over. They liked it, then showed me what they really wanted: interactive GLB models on a website, like a competitor's.

Next day

Finding another way

Within a day I worked out how the competitor did it and designed my own generation pipeline in Blender. There was no Claude Code or Codex yet, so it was all by hand. The first set of models took 15 days.

By end of Dec 2024

A studio, not a solo job

I broke the workflow into repeatable steps and hired for the repetitive ones. By the end of December I had a four-person remote studio on retainer, working on machines I paid for. That work led to an offer from Think Pro Academy.

Joining the company

Automating everything around it

I closed my small studio and built a production setup inside Think Pro. I automated the other content they made too: efficiency went up about 70% on PPTs and 95% on animated videos. GLB models became a content stream the company hadn't had before. The pipeline now runs on contract workers outside company retainers, so it adds no fixed payroll.

Then

Somewhere to put it all: the player

All that content needed a home. I built a very simple player that impressed investors and clients, and the company asked me to keep going. I delegated the whole production pipeline and gave the player my full attention.

Featured — Learno

From content player to a teacher's teaching assistant

Product owner and builder, Think Pro Academy. Live at learn.thinkpro.academy.

Where it started. Teachers needed a way to open the content we were making. That was the whole brief.

What I learned by watching. I sat in classrooms and in sessions our own teachers delivered, and watched how a class actually runs on the platform. Teachers weren't short of content. They were short of time and attention, and every extra step between them and the lesson was friction.

The goal moved from "give teachers access to content" to "make it as easy as possible to find, open and teach from it."

How I worked.

  • Every feature had a teacher behind it. At least one teacher gave feedback on each one, whether they liked it or not. Features were also agreed with the founders and stakeholders.
  • Fast loops. Building with Claude Code meant shipping about two features a day, so I could put a change in front of a teacher and learn quickly whether it mattered.
  • Cutting over adding. With shipping that cheap, the risk was a crowded product. So I kept removing things. The direction was less for the teacher to operate, not more.

What that produced.

  • Lessons that keep working when school internet doesn't. Offline sessions, plus clear loading progress.
  • Sign-in by phone number. It records who uses the player without getting in a teacher's way.
  • Parts lists teachers tick off as they go, with tile sizes large enough to read from the back of the room.
  • Plain-English activity logs, written for the people who review usage rather than for engineers.
  • School-specific accounts with each school's branding, showing only the content that school should see.

On the tech: React, 3D in the browser, and a Node backend, built with Claude Code. The technical part that mattered was the production workflow behind the content: I used Python scripts (written with ChatGPT) to drive Blender, and still modelled in Fusion 360, because AI couldn't yet produce usable 3D models.

Automation inside the company

Books: one synchronised pipeline

When the company started producing books, I saw that every stage could be automated: the content can be generated, the images can be generated, the formatting can be done in Python, and the review can be done by image-to-text models. I synchronised them into one pipeline, so a book no longer has to pass through a separate person at every stage.

PPT Splitter

PPTs are heavy, and many schools have poor internet. The splitter turns a deck into a light PDF and sends any embedded video as a separate MP4, so a lesson arrives even on a weak connection. GitHub →

Build Steps

A phone-first app that turns step photos and a parts library into build-instruction decks and parts lists, so the team doesn't lay out slides by hand. GitHub →

Why my background matters

I learned Python in 11th and 12th grade, then spent my B.Tech on 3D design and mechanical design parameters, not coding. Python is still what I understand best and can tweak, so most of my tools use it.

LLMs have made writing code roughly equal for everyone. That leaves the niche skills, like 3D design, manufacturing logic and knowing how a classroom runs, as what sets people apart. It's why I can outperform most people in this narrow area.

My own work shows it. I once wrote a full Blender script by hand. Now I build whole apps in Python, and the code I write myself keeps shrinking while the time I spend on user research and product research keeps growing.

More

Product Manager · Think Pro Academy

Feb 2026 – present · Bangalore · reports to the founders, one of 7 core team members

  • Automated teacher readiness across ~100 schools with WhatsApp, voice recognition, progress tracking and escalation.
  • Built automated quality checks for AI-generated lesson content (reading level, terminology, safety).

Before this

Led a defence surveillance-drone design to spec in 2 months against a 6-month plan (Bramer). Found 20% cost savings across 6 electric two-wheelers (Yatri). Captained a 30-person team to two national wins (Team Raftaar).

B.Tech, Mechanical Engineering, Delhi Technological University (2025).

Contact

Looking for AI product roles, remote or Bangalore. Email is the quickest way to reach me.

vaibhav.ky.2003@gmail.com GitHub Resume (PDF)