At Milya AI, I owned the development of Milya Scribe, an AI transcription platform built with Next.js, TypeScript, and Firebase that’s used by hundreds of people daily. It can process more than two hours of audio in approximately 7–8 seconds.
I migrated speech-to-text processing from Google Cloud APIs to self-hosted AI services on AWS EC2 GPU infrastructure, reducing monthly transcription costs from CAD $2,000 to $1,200. I also engineered asynchronous audio workflows that handle more than 500 transcription jobs daily.
For Milya Scribe, I built a multi-user organization feature using Node.js REST APIs, RBAC, and WebSockets, supporting 20+ organizations and 80+ users. I also created a self-service configuration experience for an AI receptionist platform that supported 8,800+ automated calls.
Previously, I developed Spring Boot APIs and optimized PostgreSQL queries at Nexowa, improving average query performance by 20%. At JVS Technologies PVT LTD, I enhanced React components and built reusable interfaces that reduced code duplication by 30%.
