Reza Ridwan

About Reza Ridwan

Full Stack AI Product and Agent BuilderI build AI products and multi-agent systems that reach over 10 million users worldwide by combining AI-human interaction understanding and AI craftsmanship.At Microsoft AI (MAI), I lead 0→1 development for Copilot across mobile and desktop surfaces (Edge, Bing, Windows). I design multi-layer eval systems (reasoning, safety, memory-grounding), prototype agent workflows, and collaborate with researchers to shape model behavior and retrieval quality. My work directly drives user growth, content intelligence, and trust.I also built 🦞 Joey (See him in action: https://diary-with-ai.com), a persistent AI operating system that runs 24/7, manages my calendar, triages my inbox, sends messages across Telegram and WhatsApp, Slack and runs 20+ automated workflows. He has his own memory architecture, security model, and graduated autonomy. Built on OpenClaw and Claude Opus. Across Microsoft, TikTok, Remitly, EA, and Samsung, I\'ve shipped products in social, fintech, media, and hardware ecosystems combining product strategy with hands-on prototyping, experimentation, and eval-driven iteration.Outside work, I\'m training for HYROX, skiing through winter, and building toward an Ironman 70.3.Let\'s connect if you\'re interested in AI agents, personal OS systems, eval frameworks, or performance training.

Experience

1. Senior Product Manager

   Microsoft AI

   Mar 2025 — Present · Redmond, WA, US

   Leading AI product development for Copilot Consumer across Copilot, Bing, Edge, and Windows surfaces — building the next generation of curated companion and discovery experiences for 100M+ users.Focus areas:→ Built 0→1→N multi-agent orchestration with tool-calling patterns for Copilot Finance vertical. Designed scalable factory pattern with structured tool calls, state management, and quality gates — now extensible across verticals→ Created eval frameworks (IQ/EQ/AQ + Responsible AI) with human-AI judge calibration, improving pass rates from 30% to 90% across groundedness, safety alignment, and hallucination detection→ Drove external partner integrations via tool-calling APIs for grounded financial data. Managing vendor performance across quality, cost, and UX tradeoffs→ Prototyping full-stack (backend + frontend) with Claude Code and Replit — shipping prompts and agent architectures directly to production→ Collaborating with researchers and engineering on model selection, prompt pipelines, inference constraints, and precision/recall improvements for content understanding

Education

• Carnegie Mellon University - Tepper School of Business

  Master of Business Administration - MBA, Technology, Strategy, Data Analytics, Product Management

  2017 — 2019

• Institute of Business Administration, University of Dhaka

  Bachelor's degree, Finance, Consumer Psychology

  2010 — 2013

• Notre Dame College

  Physics, Chemistry, Mathematics, Biology