Nikhilesh Gawhale

– Designed and deployed agentic Al workflows using LangGraph, AutoGen, and CrewAl to automate repetitive tasks and improve operational efficiency by 20%. – Built Python services with FastAPI and WebSockets to support real-time automation and content processing workflows. – Contributed to full-stack delivery by integrating backend logic with user-facing workflow interfaces and internal tooling. – Implemented prompt engineering and evaluation workflows to improve output quality, safety, and reliability in production systems. – Added monitoring and logging to track performance, errors, and model behavior in production environments.

– Built and deployed a document analysis and summarisation system for financial and business reports using LLAMA 2 and RAG pipelines. – Developed retrieval pipelines over 10K+ documents using Milvus to ground LLM outputs and reduce hallucinations. – Designed prompt engineering strategies that improved response accuracy and consistency by 18%. – Worked with cross-functional stakeholders to define Al use cases aligned with business goals and data governance requirements. – Optimized GPU workloads on Kubernetes, reducing compute cost by 30% while maintaining performance.

– Built ML-powered automation tools to streamline QA and operational workflows, improving efficiency by up to 45%. – Developed predictive pipelines and monitoring tools to reduce defect rates and improve system reliability. – Collaborated with product and QA teams to translate operational needs into scalable automation solutions. – Designed lightweight data validation and exception-handling checks to reduce manual rework and improve workflow consistency. – Supported rollout of workflow improvements by working directly with end users and incorporating feedback into implementation.