AI Systems Engineer | LLM Applications, RAG & Backend Systems | Building Real World AI Pipelines

Ucchishth Singh  • Peerlist

<p>Built a configurable RAG experimentation framework to evaluate chunking strategies, retrieval depth, and prompt variations. Implemented FAISS based vector search and embedding-driven evaluation metrics such as answer similarity and retrieval coverage. Designed grid search pipelines to identify optimal RAG configurations.</p>

AutoRAG Optimizer | RAG Evaluation Framework

Built a temporal aware LLM routing system that detects time sensitive queries and automatically triggers live web search to improve response accuracy. Designed a backend pipeline with async retrieval, content extraction, and relevance ranking to generate grounded answers with source citations. Achieved ~80% routing accuracy on mixed query evaluation, improving reliability of LLM responses beyond knowledge cutoff limitations.

AutoSearch LLM | Temporal Aware LLM Routing S

Built an AI-powered deployment intelligence system that predicts high risk code deployments before production. Designed a FastAPI based backend that analyzes commit diffs and correlates them with historical incident patterns to generate contextual risk scores. Implemented automated decision logic to allow or block deployments with structured, explainable outputs.