60 FPS interfaces with GSAP motion, design tokens, and strict type safety.
Front
end
- React
- Next.js
- TypeScript
- JavaScript (ES6+)
- Tailwind CSS
- GSAP
- shadcn/ui
- Framer Motion
- Three.js
- React Hook Form
- Zod
I build real-time AI systems — agents in live calls, webhook-driven pipelines, async job queues — and ship them to production, not just to a portfolio.
Nothing here is a line on a résumé I have not shipped with. Every tool below appears in at least one project further down the page.
60 FPS interfaces with GSAP motion, design tokens, and strict type safety.
High-throughput schemas, tRPC contract safety, connection pools, and Redis caching.
Sub-second WebRTC pipelines, OpenAI Realtime agent streaming, and multi-instance WebSockets.
Decoupled webhook ingestion with Inngest job queues, QStash tasks, and Docker containers.
Real-time AI voice and video call infrastructure. Interactive agents stream into live calls over WebRTC, and every call ends with an AI review of its own transcript.
Stream webhooks trigger all real-time events. Transcript summarization is decoupled from the webhook response by an Inngest job queue, so the handler returns immediately and the heavy work happens off the request path.
I am a full-stack developer specializing in building high-performance systems and interactive digital products. From low-latency AI integrations and real-time WebSocket pipelines to pixel-perfect client interfaces, I construct robust software with React, Next.js, GSAP, and TypeScript.
Stream webhooks trigger all real-time events. Post-call transcript summarization is decoupled from the webhook response via an Inngest job queue — the handler acknowledges fast, the AI review runs as a background job.
A custom HTTP server running Socket.io with the Redis adapter, so notifications and messaging survive multi-instance scaling instead of breaking the moment a second container starts. Dual deployment config — Vercel and Fly.io via Docker.
Redis-backed caching and rate limiting in front of the database, QStash for serverless async job queuing, and single-session enforcement on admin accounts. Full admin surface: inventory, order pipeline, CRM, financial analytics.
Digivixo is live and selling $1,500–$2,200/month plans to real accounting firms. The Soft Biz project was a complete solo backend build for an AI-led interview platform — client and product withheld under NDA.

Webhook-driven architecture — Stream webhooks trigger all real-time events. Async transcript summarization decoupled from webhook response via Inngest job queue.

Custom HTTP server with Socket.io + Redis adapter enabling multi-instance WebSocket scaling. Dual deployment config — Vercel and Fly.io via Docker.

Redis-backed caching and rate limiting. QStash for serverless async job queuing. Single-session enforcement for admin accounts.

Live client product currently selling $1,500–$2,200/month plans to real customers.

Client and product name withheld under NDA — complete solo backend build.
I architect and ship high-performance full-stack applications and real-time AI systems — from sub-second agent streams over WebRTC to distributed background queues. Open to engineering roles, client contracts, and technically ambitious builds (no system is too complex, and no idea is too raw).
Download Resume