MP Thaker

I ship AI to production, not PowerPoint.

Prevented $400K+ in churn. Built AI platform that cut PM busywork 80%. Built 3 companies with 1 successful exit.

work

lovingly · sep 2024–present

product manager

mandate: Turn a 15-year-old e-commerce platform into an AI-native operation without breaking P&L or trust.

what I built:
  • AI transformation platform: Multi-agent system reducing PM overhead by 80%
  • Re-platforming initiative: SHOP and CHECKOUT module migration to reduce unplanned work 50% quarterly
  • Support automation: AI-powered ticket triage and response system
results (tracked with RevOps/Finance):
  • $400K+ annual churn prevented through early detection and intervention
  • $6M revenue opportunity identified through AI-enhanced analysis
  • 80% reduction in repetitive PM task time, 27% team velocity improvement
  • Cross-functional impact: 10+ engineers, designers, marketers benefiting from automation

stack: OpenAI / Anthropic abstraction, internal tools, vector store, offline eval harness (precision/latency/cost), red-team tests

2016–2024

founder / head of product

Built 3 companies with 1 successful exit (Subziwalla)

  • Subziwalla: 70% retention (2.3x industry), 180% YoY growth, 100:1 LTV/CAC ratio at peak
  • Panso: IKEA partnership secured, 65% email open rates (3x industry), 319,950 footfall achieved

systems & experiments

toolkit

View on GitHub

frameworks

ai usage excellence guide

responsible use, eval harness patterns, safety gates

product development master

discovery → launch with decision gates & metrics

structured dialogue system

meeting → decisions → artifacts

sprint metrics framework

flow, velocity, and early-warning signals

decision-driven development

every work item tied to a single decision

pm automations

morning status generator

pulls JIRA/GitHub/Calendar → daily brief

sprint report generator

velocity trends, blockers, health scores

templates

prd template

prefilled metrics, rollout, risk checklist

jira bug

repro, impact, verification matrix

podcast

pm in the pm

real stories from shipping AI: multi-agent systems, evals, costs, and the messy middle. 10–15 minutes, one takeaway per episode.

about

the unconventional path

After building 3 companies (1 successful exit), I stepped into a hands-on AI PM role. Not for title, for reps — because the job changed, and I wanted my hands on the new tools.

what makes me useful

I've been the exec asking "should we?" and the PM proving "here's how." I build reusable AI platforms that teams can ship on, with the guardrails Legal/Sec can live with and numbers Finance will sign.

how I work

  • ship learning systems: 80% in prod beats 99% in staging
  • platform > project: every solution should make the next one faster
  • governance at speed: partner with Legal/Sec early to maintain velocity

the human

Dad in Rahway, NJ. ADHD means hyperfocus is my superpower (and timers are my friend).