Engineered AI-native radiology reporting workflows using NestJS, TypeScript, Next.js, Prisma, and pg-vector in a Turbo monorepo architecture.
Integrated Soniox Speech-to-Text API for real-time medical dictation transcription with streaming updates and post-processing pipelines.
Built RAG (Retrieval-Augmented Generation) pipelines with vector retrieval to improve report accuracy using domain-specific medical context and historical templates.
Implemented automated report generation, formatting, and structured sectioning from raw ASR transcripts using LLM-based processing.
Developed agentic workflows using LangGraph pipelines for intelligent model routing, report refinement, and workflow automation.
Improved transcription quality through VAD tuning, endpointing calibration, and noise-handling improvements.
NestJS
Next.js
RAG + Vector DB
LLM Workflows
LangGraph
Soniox STT
Daily Wellness AI
Backend & DevOps Developer
📅 Internship Period: February 2025 – July 2025
Built video streaming platform with EC2 + S3 + Nginx.
Integrated Stripe with webhook-based emails.
Connected AI services to generate nutrition plans.
Implemented CI/CD via GitHub Actions.
Used Next.js, TypeScript, and NextAuth (Google OAuth).