Krishna Vishesh Katta

About

Early-stage product engineer who built an AI persona chat system end to end at Super Intro, spanning Next.js frontend, GraphQL/real-time backend, retrieval memory, and LLM-based matching. They combine strong TypeScript rigor with practical AI systems design, and cite measurable impact including ~40% engagement growth, ~30% recall improvement, and lower LLM costs in production.

Key Strengths

• Owned full-stack AI feature end to end from UI to database and model calls
• Built real-time persona chat with memory and live presence/typing architecture
• Strong TypeScript discipline with shared contracts across frontend, backend, and sockets
• Designed multi-stage AI matching pipeline using embeddings plus LLM reranking
• Implemented hybrid RAG memory with vector search and BM25 fallback
• Improved recall accuracy by about 30%
• Increased user engagement by about 40%
• Reduced duplicate LLM calls and lowered cost per conversation with Redis caching and idempotency
• Production-minded AI engineering with schema validation, fallback handling, and regeneration loops
• Built offline evals and production tracing to catch regressions in retrieval and latency
• Effective in ambiguous early-stage environments through rapid prototyping and iterative alignment

Experience

Education

Skills

Languages