Jaskirat Singh

Leading the AI charter and creating India’s first full stack AI companions for Lifestyle.

1) Chatbot using RAG & LLMs: Developed a Retrieval-Augmented Generation (RAG) based chatbot using large language models to streamline internal document search. The chatbot enables teams to upload and query documents like contract agreements and technical documents, eliminating the need for manual keyword searches. 2) Stress Testing Credit Risk under Macroeconomic Scenarios: Built Monte Carlo simulation models to analyze how credit risk metrics behave under various macroeconomic stress scenarios. The analysis focused on understanding non-linear impacts on Expected Credit Loss (ECL) and contributed to key insights in the bank’s annual disclosures. 3) Project Finance: Worked on assessing the creditworthiness of the bank's project finance portfolio by analyzing projected cash flows and evaluating risk grades of borrowers. Also contributed to refining model parameters (MEV) used in risk evaluations. 4) Monitoring Credit Risk Deterioration: Supported the ongoing tracking of clients flagged for significant increase in credit risk (Stage 2). Monitored changes in credit exposure (EAD), analyzed migration between credit stages, and contributed to lifetime ECL calculations for early warning and provisioning.

Project 1: Generative AI – Private LLM Development Worked on building local Large Language Models (LLMs) on private infrastructure to ensure complete data confidentiality. The solution enabled the use of generative AI capabilities without exposing sensitive data to external servers, aligning with privacy and compliance requirements Project 2: Analysis of Expected Credit Loss (ECL) Applied stress testing methodologies on macroeconomic variables to assess their impact on Expected Credit Loss across various bank portfolios. The analysis supported risk management and capital planning by quantifying potential credit deterioration under adverse economic scenarios..

Project 1: Power Loading Optimization Implemented and compared multiple optimization algorithms to maximize power loading efficiency of a helicopter, while adhering to rotor blade performance and structural constraints. The project aimed at improving flight performance under complex engineering limitations. Project 2: ODOR – Chemical Sensing Device Collaborated in a multidisciplinary team to design and prototype ODOR, a chemical-sensing device using olfactory receptor technology. The device was envisioned for applications in testing laboratories and space stations, where real-time chemical detection is critical.

Received Convocation Award for Best Project Work in the Mathematics Department at IIT Kanpur Presented work on Private GPTs and RAG at IndoML 2023 held at IIT Bombay Delivered a conference talk on Non-Monotonic Logic at the International Conference of Logic and its Applications 2023 at IIT Indore Secured AIR 651 in JEE Advanced 2019