Skip to content
Nitesh Tiwari
Back to home

Work

Product decisions, with the evidence behind them.

Four product stories in depth, then the smaller calls as short snapshots, including the bets that were stopped, then the AI Lab.

Four stories

  1. 01  Strategic trade-offs

    Faster answers without scaling support

    Edfora · EdTech · myPAT · Senior Product Manager · 2023–2026

    Senior PM × Strategy × Consumer × AI judgment

    An AI auto-resolver, a tutor marketplace, or a hybrid. I evaluated them on RICE and unit economics; the hybrid was piloted for two weeks with 10,000 JEE students behind a 90% accuracy circuit-breaker, and the AI auto-resolver was evaluated but not shipped as the first solution.

    1. Signal

      Doubt resolution approached 24 hours at peak exam preparation.

    2. Decision

      Guided hints, verified peer solutions and SME escalation, behind a 90% accuracy circuit-breaker. The AI auto-resolver was evaluated but not shipped as the first solution.

    3. Trade-off

      Scale against academic integrity: take the repetitive volume off faculty without letting unverified answers through.

    4. Outcome

  2. 02  Activation

    Fixing the path to first value

    Witzeal Technologies · Real-money gaming · Product Manager · 2022–2023

    Growth × Activation × Retention × Experimentation

    Read as a retention problem, it pointed to reminders and rewards. The funnel said activation. Five changes to the first 60 seconds, tested against a 30% control, and a clear account of what the test could and couldn't isolate.

    1. Signal

      Only 12% of new users played a game on day one.

    2. Decision

      Five changes to the first 60 seconds, tested against a 30% control.

    3. Trade-off

      Shipping the five changes as one bundle was faster; the price was attribution, since one of them was ₹15 of free games.

    4. Outcome

  3. 03  Personalization

    Personalizing the learning path

    Edfora · EdTech · Adaptive Practice · Senior Product Manager · 2023–2026

    Personalization × Product logic × Technical depth

    Three ways to fix difficulty fit, one chosen: a 3PL IRT engine that estimates each learner's ability and matches the question to it.

    1. Signal

      Every learner was getting the same next question.

    2. Decision

      A 3PL IRT engine that estimates each learner's ability and matches the question to it.

    3. Trade-off

      Fit against explainability: every question needs calibrated parameters, and a new learner's first questions are the least certain.

    4. Outcome

  4. 04  Behavioural systems

    Designing behavioural loops across students and faculty

    Edfora · EdTech · Glorifire / Stakeholder · Senior Product Manager · 2023–2026

    Consumer engagement × Behavioural systems

    A configurable behavioural engagement system across students and faculty: rewards tied to product actions, visible progression and recognition, and analytics for faculty, built as one loop rather than isolated point mechanics.

    1. Signal

      DAU had been flat for two straight months.

    2. Decision

      A configurable behavioural loop built around product actions, rather than isolated point mechanics, for students and faculty alike.

    3. Trade-off

      Motivate students without looking gimmicky to the teachers who had flagged early prototypes as “too game-y”.

    4. Evidence

      Behaviour → reward → progression → recognition → analytics, configurable per action. DAU and session time were reported up after launch; no retention or business outcome is claimed, and no change is causally attributed.

Decision library Short snapshots: signal, decision, outcome or learning.

Read every decision
AI Lab

AI product judgment: problem first, model second.

Strong AI product judgment; production evidence in progress. No evaluation has been run.

Inside the AI Lab