Abhitesh Bhardwaj

AI Engineer specializing in low-latency LLM systems, Retrieval-Augmented Generation (RAG), and multi-agent orchestration. Experienced in building production-grade semantic search and real-time voice AI systems with a focus on scalability, reliability, and efficient inference. Strong foundation in algorithmic problem solving.

AI Engineer at Inscripta AI (2024-01 – Present)

AI Engineer at Inscripta AI working on LLM evaluation, semantic search, and RAG systems.

• Evaluated multi-provider LLM performance (OpenAI, Gemini, Claude) on real production queries, analyzing hallucinations, tool reliability, and response consistency to guide model selection.
• Designed structured evaluation workflows and benchmarking pipelines enabling reproducible testing and systematic validation of GenAI systems.
• Identified hallucination patterns linked to large tool outputs and validated system fixes, ensuring reliable production deployment.
• Built a hybrid semantic search system combining dense embeddings and keyword-based retrieval, achieving over 80% search accuracy.
• Developed a production-grade RAG pipeline with custom parsers and retrieval optimization, improving answer accuracy and response relevance across client datasets.
• Built a context-aware QA system using RAG pipelines and optimized retrieval strategies for efficient information access across large-scale datasets.
• Built the frontend for a multi-agent planning system with real-time streaming, session management, and agent trace visualization for debugging and monitoring.

B.Tech (Hons.) in Computer Science and Engineering – Lovely Professional University (2020-07 – 2024-08)