Bengaluru·Architecting LLM-native UI primitives and AI-augmented developer workflows for the Indus app.
- 01.01Indus app, 0→launch in 2 weeksArchitected and shipped the Indus app end-to-end. Launched at 4.8 on App Store, 4.6 on Play Store.
- 01.02Crash-free reliability at 99.99%Drove crash rate from 16% → 0.01% via end-to-end observability, hardened render boundaries, rationalised native-bridge contracts.
- 01.03OEM partnership with HMD Global (Nokia)Closed a strategic deal to pre-bundle Indus across HMD's device portfolio — firmware-level organic distribution.
- 01.04MCP UI widget frameworkDesigned and delivered an end-to-end framework enabling LLMs to render rich, agentic, multi-modal interfaces inline. Demoed live with Swiggy at the Razorpay FTX 2026 keynote.
- 01.053–4× faster feature cycles via AI workflowsOrchestrated AI-augmented developer workflows across the team — Cursor, custom Claude Code agents, MCP tool servers, prompt-engineered scaffolding, skill-driven automations.
- 01.06Platformised the frontend stackSet direction for shared runtime contracts, schema-driven UI, unified rendering primitives. Elevated the design system with refined motion and craft-led primitives.