Every component
must earn its place.

AI governance platform — a controlled desktop browser + analytics dashboard helping organisations measure and govern LLM usage. Core team. CTO collaboration.

• Built the core data capture engine using Chrome DevTools Protocol — LLM conversations intercepted, processed through a four-stage classification pipeline, stored to MongoDB
• Engineered tabbed session architecture with isolated per-tab session partitions and full state preservation on switch
• Cut release cycle from 30–45 min to 13 min via shell scripts + GitHub Actions CI/CD
• Config-driven AI provider runtime — ChatGPT, Claude, Meta AI onboarded via JSON config; MVP shipped before deadline

SaaS spend intelligence platform. Owned both desktop and browser agents — silent background processes streaming ~1M events/day to AWS Kinesis across 20 enterprise orgs including monday.com and Saltpay.

• Replaced manual OTP login with device UUID + hashed org token via Auth0 — silent MDM deployment to thousands of devices simultaneously across Jamf, JumpCloud, Kandji, Intune
• Cut customer MTTR from ~6 hours to 15–30 min by building a diagnostics window with exportable logs and a cache-flushing force-restart that re-authenticated via Auth0 cleanly
• Cross-platform agent delivery — macOS, Windows, Ubuntu; PouchDB as local event store with Kinesis stream sync

Hyper-local talent discovery — find a human teacher near you for any skill, via maps UI. Built solo from scratch. Selected from 700+ applicants at iCreate Ahmedabad, 2019.

• Built entire platform solo — PHP/MySQL/jQuery backend, DB, and frontend — later architecting two full migrations to REST API and React PWA
• Operated as sole technical decision-maker in a zero-resource environment; hired and mentored three interns
• Selected from 700+ applicants at iCreate Ahmedabad — presented to industry leaders, underwent structured product mentorship