Jul 2025 – Present · 11 mos · Bengaluru · Hybrid

• Exploring the Physical AI domain
• Build and optimised RAG application in production

Oct 2024 – May 2025 · 7 mos · Bengaluru, Karnataka, India · Hybrid

• 1. Built a multi-agentic AI system tailored for industry-specific problem solving, coordinating specialized agents via a supervisor-based architecture and utilizing in-memory state management and external tools.
• 2. Designed and optimized a Retrieval-Augmented Generation (RAG) evaluation pipeline, benchmarking open-source RAG systems to enhance response quality, contextual accuracy, and system scalability.
• 3. Integrated the RAGAS module to systematically evaluate generation quality across metrics such as factual correctness, contextual grounding, and hallucination rates.
• 4. Implemented autonomous self-discovery agents using frameworks capable of conducting deep research tasks such as literature review, regulatory analysis, and risk assessment.
• 5. Leveraged advanced LLM-powered research tools (OpenAI Deep Research, Perplexity, Google Gemini) to automate data collection, analysis, and technical report generation.
• 6. Explored LLM benchmarking using frameworks like OpenCompass and LM Evaluation Harness, analyzing prompt strategies, model performance, and robustness across a range of industrial NLP tasks.