Manasa Reddy Nagendla

About

Software engineer at Procter & Gamble focused on warehouse/operations systems, building near-real-time order/inventory visibility using Java/Spring Boot, React, Kafka, PostgreSQL, and Redis with measurable latency and load-time gains. Also shipped internal LLM/RAG knowledge assistants grounded in company runbooks and workflows, implementing guardrails and an evaluation loop that drove concrete retrieval improvements (document chunking) and regression prevention.

Key Strengths

• Owned end-to-end internal dashboard feature (APIs + React UI) for warehouse order monitoring
• Reduced cross-system latency ~60% by replacing API polling with Kafka event-driven streaming
• Improved API response time from ~2.4s to <900ms via SQL optimization and Redis caching
• Improved peak-time dashboard load time ~35% based on operations-team feedback
• Designed RAG-based internal knowledge assistant grounded in internal docs with guardrails (restricted to retrieved sources, low-confidence fallback, interaction logging)
• Built LLM evaluation loop using real support-ticket/Slack questions; tracked retrieval relevance and hallucination/unsupported info
• Improved RAG accuracy by changing ingestion to smaller document chunks after evals showed poor retrieval on large docs
• Built reliable, observable event pipeline with normalization, retries, DLQ, and monitoring (CloudWatch/ELK) while balancing latency vs cost via batching

Experience

Education

Skills

Languages