AI Engineering Specialist at Setu Business School Pvt Ltd (2025-04 – Present)
- Designed and built a LangGraph based multi-agent assessment platform using asynchronous Python and Pydantic-based state management and Langsmith for observability, reducing SME dependency by ~90% and manual assessment-creation effort by ~85%.
- Implemented LLM-as-a-Judge evaluation and verifier workflows for competency, sub-competency, and question quality; the pilot was demoed and tested by enterprise clients including EY, InterGlobe and MOVIN, with average generation cost of ~$0.025 per question and ~$0.39 per competency/sub-competency generation cycle.
- Developed a RAG-powered document intelligence prototype for EY, enabling users to query complex PDFs and retrieve grounded, page-cited answers; implemented semantic chunking and hybrid FAISS + BM25 retrieval using Unstructured, LangChain and Gemini 2.0 Flash, reducing manual document review effort by 30%.
AI Engineering Specialist at Setu Business School Pvt Ltd (2022-10 – 2025-03)
- Built loan default model for a third party client to correctly identify default loan applications with 67% F1 score in logistic regression using feast, mlflow, sklearn, dvc, google artifact registry.
- Trained 500+ corporate professionals from Big4 firms, banks and large MNCs in Data Engineering, mlOps.
Data Science Consultant at SunShell Power (2021-04 – 2022-10)
- Worked with project and HR team on designing, implementing the centralised database for project and employee management, NLP-based workforce analysis using Python, Pandas, scikit-learn
