Software Engineer — Backend & AI Integration @ nference · Bangalore, India
🔗 devratapuri.github.io ·
LinkedIn ·
Email
- 🛠️ 3+ years building production backend systems and full-stack platforms at scale — Go, Node.js, TypeScript, Python, React across microservices, REST APIs, PostgreSQL, Redis, and AWS.
- 🤖 Currently building LLM-powered and agentic workflows in production at nference — prompt chains for automated extraction/summarisation, agentic pipelines connecting LLM inference with internal data APIs.
- 📈 Impact so far: 40% MTTR reduction, 30% faster release cycles, 20% p99 latency improvement.
- 🌱 Deepening AI/LLM engineering fundamentals — transformers, RAG, agents, and evaluation — end to end.
- AI Agent Evaluation Pipeline (Go, PostgreSQL, Docker, Anthropic API) — production-grade pipeline ingesting 1000+ agent conversations/minute via concurrent evaluators (heuristic, tool-call accuracy, LLM-as-judge, multi-turn coherence); self-updating feedback loop calibrating LLM-as-judge scores against human annotations.
- Rust Async Real-Time Chat System (Rust, WebSockets, Redis, Tokio, Docker, Nginx) — high-concurrency async WebSocket server handling 1K+ concurrent connections with Redis pub/sub fanout, room-based routing, and fault-tolerant reconnection.
- API Documentation Generator (Go, OpenAI API, Swagger/OpenAPI) — CLI tool auto-generating and enriching OpenAPI specs for Go/Node.js services; cut developer onboarding time by 25%.
