About
I build products at the intersection of growth, consumer experience and AI.
Growth × Consumer × AI
I started by writing the product: four-plus years as the only Android developer on a marketplace app, then release management for enterprise web applications. That order still shapes how I work: I scope with engineering, not around it.
Gaming at Baazi Games and Witzeal taught me how users behave and how growth and monetization work. At Edfora I worked on adaptive practice, doubt resolution and engagement systems. AI is my current direction, through an independent prototype rather than a role.
Growth
Activation, retention and experimentation.
Consumer
Experiences designed around user behaviour.
EdTech
Personalization and learning products at scale.
AI
Practical, responsible AI product judgment.
What each role taught me, and what I build now
- 01 · 2014–2018
Built
Direct Create
Writing the app from scratch taught me what a feature really costs, and that one well-scoped app can beat three tailored ones.
Sole Android developer: built the app from scratch, crash rate down ~30%.
- 02 · 2019
Shipped
PwC India
Release management taught me that a product is only as good as the way it reaches people: process, sequencing, four teams in step.
Standardized releases for enterprise web applications across four distributed teams.
- 03 · 2019–2022
Measured
Baazi Games
Gaming taught me how users behave, and how to measure it. Twenty-plus experiments, many of them inconclusive, taught me to scope bets smaller.
Experimentation, segmentation and risk across PokerBaazi, Lagai Khai and FanBlaze.
- 04 · 2022–2023
Grew
Witzeal Technologies
Growth taught me to find where users drop before they reach value, and to fix the product before paying them to come back.
Onboarding, bonus economics and lifecycle messaging for a real-money gaming platform.
- 05 · 2023–2026
Personalized
Edfora
EdTech taught me to adapt a product to each person with a statistical model, and to be honest about what a before/after can and can't prove.
Adaptive practice, doubt resolution and engagement systems. Edfora's products reached 100K+ learners overall.
- Now · independent, not a role
AI
Independent prototype · AI Learner Diagnostic
AI is the next application of the same discipline: a clear user problem, a measured outcome, and an evaluation agreed before the demo.
AI Learner Diagnostic: an independent prototype, evaluation designed before any model work.
The engineering behind the product calls
The drop wasn't motivation. It sat in signup and OTP verification, before the first game.
- Built
- 4+ years as the sole Android developer: Firebase real-time chat and file sharing, OAuth 2.0, encrypted local storage; crash rate down ~30%.
- Diagnosed
- Read the OTP API's success and failure rates and delivery time alongside the funnel, which moved the fix from campaigns to signup.
- Specified
- Wrote the PRD and adaptive product logic for a 3PL IRT engine: inputs, the ability-update loop and its edge cases. The model was built with engineering.
- Measured
- Experiments from hypothesis and sample size to significance; a 30/70 controlled rollout; a pilot read against a holdout.
- Prototyped
- An AI diagnostic with a deterministic baseline, an output schema and an evaluation harness, before any model work.
How I work
Four principles, each with a place in the work where it shows.
01
Where it showsDiagnose before building.
Find where users actually drop, and why, before choosing what to build.
A 12.2% Day-7 number pointed to reminders and rewards. The funnel showed only 12% of new users played a game on day one: an activation problem, with a different fix.
Witzeal · Diagnosis
02
Where it showsOne outcome decides. The rest explain.
Pick one outcome tied to user value, and treat activity metrics as diagnosis rather than success.
At Edfora, assignment completion was the success measure for adaptive practice, with practice drop-off defined as the second signal.
Edfora · Hypothesis
03
Where it showsExperiments can fail.
Experiments are not successful because they win. They are successful because they reduce uncertainty.
20+ A/B tests at Baazi Games, a fair number inconclusive or negative, and they changed how later tests were scoped. FanBlaze live scores were sunset at under 6% usage.
Decisions · Experimentation, FanBlaze
04
Where it showsAI needs evaluation.
Agree the rubric, the failure modes and the launch gate before anyone argues about the demo.
In doubt resolution, the quality bar was a number before launch: the shipped hybrid ran behind a 90% accuracy circuit-breaker with a rollback, and >90% accuracy was maintained by SME sampling. The AI auto-resolver was evaluated and not shipped as the first solution.
Edfora · Doubt resolution
Experience
- Jul 2023 – Jul 2026
Gurugram
Edfora · Senior Product Manager
Across Edfora's products and systems, including myPAT, Glorifire, Stakeholder, Glorifire Ops, Adaptive Practice, myAdvisor and myPlan: roadmap and prioritization for learning and engagement experiences on web and mobile, with engineering, design, content and business teams.
- 3PL IRT-based adaptive assignments: across a 2-year academic-cycle dataset, completion was 18% on the static path and 45% after (not attributed to the engine alone)
- Early prototypes were flagged by teachers as “too game-y”. DAU was reported up 12–15% after the quiz and gamification layer. Average session time was reported up ~15%.
- Doubt resolution on myPAT and Glorifire, owned end to end: evaluated an AI auto-resolver, a tutor marketplace and a hybrid on RICE and unit economics; 2-week, 10,000-student pilot behind a 90% accuracy circuit-breaker; from a ~24 h peak bottleneck (reported) to a <15 min median for repetitive doubts (measured); +18% relative uplift in D14 retention, measured via a pilot holdout / A-B cohort; >90% accuracy maintained; ~60% support-cost reduction (derived)
- Real-time engagement dashboards for faculty replaced a monthly spreadsheet pull
- Interviews and usability tests with students and faculty fed a RICE-based roadmap; 5+ features shipped across web and mobile. Edfora's products reached 100K+ learners overall
- May 2022 – Mar 2023
Gurugram
Witzeal Technologies · Product Manager
Growth, onboarding, monetization and lifecycle for a real-money gaming platform.
- Onboarding redesign, shipped as one bundle of five changes: Day-7 retention 12.2% (control) vs 25.4% (treatment), +13.2 pp, in a 30/70 controlled rollout on ~50K users over ~3 weeks. D0 gameplay was 12% before the redesign and 33% after; that comparison basis isn't recorded.
- Experimentation roadmap across pricing and reward loops, each change written as a hypothesis and run as an A/B test
- Bonus allocation by expected ROI per segment: bonus and discount spend down ~20%, retention held
- Lifecycle messaging (push, in-app, email) moved from one blast to segmented cohorts
- Jun 2019 – May 2022
New Delhi
Baazi Games · Product Manager
Experimentation, segmentation and risk across a multi-game platform: PokerBaazi, Lagai Khai and FanBlaze.
- 20+ A/B tests end to end (hypothesis, sample size, significance); a fair number came back inconclusive or negative and reshaped later scoping
- Behavioral clustering replaced one default journey with journeys by segment: session duration reported up ~35% and retention up ~25% (method, window and unit not recorded)
- Rules-based anomaly detection for fraudulent transactions: fraud losses reported down ~18% (method and window not recorded)
- Jan 2019 – Jun 2019
Gurgaon
PwC India · Program & Release Manager
Program and release management for enterprise web applications, including EwayBill and an LMS.
- Standardized the release process for an enterprise web application
- Coordinated four distributed teams across development, QA, UAT and deployment
- May 2014 – Dec 2018
Gurgaon
Direct Create · Android Developer
Sole Android developer on a collaboration platform for the handmade industry, working directly with the CEO and CTO on what to build.
- Proposed one app with role selection instead of separate maker, buyer and designer apps
- Real-time chat and file sharing on Firebase; crash rate down ~30% through better state handling and testing
- OAuth 2.0 and encrypted local storage; 4.6+ Play Store rating kept through iterative UX fixes
Hiring for a product role? Let's talk.
Open to Senior Product Manager and Product Manager roles, especially in growth, consumer products and AI. Delhi NCR · open to Mumbai.