Technical Product Manager.

I started by designing Pakistan's first indigenous satellite. Now I ship AI products, and I build them myself when I can: specs, code, deployment.

I hold a B.E. in Aerospace Engineering from the Institute of Space Technology, Islamabad, and two Master's degrees in Mechanical Engineering — one from NUST (Computational Mechanics, completed part-time while working full-time at Pakistan's national space agency) and one from the University of Massachusetts Amherst, where I studied as a two-time Fulbright Scholar (GPA 3.80/4.0). I'm also a Google Certified Project Manager and hold IBM's Generative AI for Product Managers certification.

• SigmaSight B2B SaaS · AI

  Turning AI-accelerated development from a QA bottleneck into an automated pipeline

  Pilot proof of concept complete, now in pilot

  As AI-assisted coding sharply increased ticket throughput, QA became the constraint on how fast the team could ship.

  1. AI-generated acceptance criteria at ticket creation
  2. Playwright tests auto-run on staging
  3. Results posted, QA and PM tagged
• SigmaSight DevOps · Automation

  Automating release tracking for trunk-based deployment

  0 manual release updates or Dev→Staging ticket moves

  After the team moved to trunk-based deployment alongside AI-accelerated development, manual Jira release management stopped scaling.

  1. Jira webhook
  2. GitHub Actions extracts release from tag
  3. Release created, fix version on every ticket
  In active production use
• Transconomy AI · Infrastructure

  Closing and delivering the largest deal in Transconomy's history

  1st time the company went revenue-positive

  Led the Richland County engagement end-to-end, from the first technical sales call to final delivery.

  1. Technical sales call
  2. Proposal drafting
  3. Deliverable breakdown
  4. Execution
• Transconomy Mobile Product

  Turning a customer complaint into a field app

  40% less time spent on data collection, reported by collectors after launch

  Customers wanted to use their own GoPros, and our existing collection app didn't fit how they worked in the field.

  1. Draw regions on a map
  2. Guided routes with live coverage
  3. GoPro records continuously
  4. Geo-tagged voice notes
  Rolled out mid-project on Richland County
• Transconomy Computer Vision

  Estimating pothole size from ordinary 2D road imagery

  2 new offerings: PCI and PASER pavement ratings

  Road authorities need pothole dimensions to rate pavement condition, but our pipeline only captured standard 2D video, with no depth sensors.

  1. 2D street imagery
  2. Pothole detection
  3. Size from image geometry
  4. PCI and PASER ratings
  No new hardware, using data customers already collected

• Consumer AI (in stealth)

  Building and shipping a consumer AI mobile app solo, from spec to production

  ~2× free-tier AI quota (about 500 to 1,000 requests a day) through automatic model fallback

  I'm building a consumer AI app on my own and run it like a real product: every feature starts as a written spec before any code is written.

  1. Written spec
  2. Flutter mobile app
  3. FastAPI + Supabase auth
  4. Gemini with model fallback
  5. Google Cloud + Docker
  In production on Google Cloud · walkthrough on request
• AI Voice Journaling (pre-launch)

  Designing an AI product to run at $0 per entry

  $0 per entry, vs. about $0.02 on the production setup

  A voice journaling app I'm building solo: users speak, and AI turns each entry into a structured, searchable summary.

  1. Voice recording
  2. Speech-to-text
  3. LLM summary
  4. Embeddings for search
  5. Live results in app
  Working MVP · 18 of 23 stories done Read the case study →