- Evaluating Compound AI/LLM Systems through Behaviors, Not Benchmarks
- HOLMES: Hyper-Relational Knowledge Graphs for Multi-Hop Question Answering using LLMs
- C2FDrone: Coarse-to-Fine Drone-to-Drone Detection using Vision Transformer Networks
- Real Time Estimation of Weighted Shapley Values
- Interpretable Model Drift Detection
- Feature-Interpretable Real Concept Drift Detection
- Instance-wise Causal Feature Selection for Model Interpretation
- Forest Ground Vegetation Cover Estimation
Research Experience
ML Researcher at Fujitsu Research. Interned at Amazon Science (India). Worked as a Research Assistant in the CSE department, IIT Hyderabad, during Master's.
Education
MTech (by Research) in CS from IIT Hyderabad; BTech in ECE from NIT Rourkela. Advised by Prof. Vineeth N Balasubramanian during Master's thesis. Grateful for the guidance provided by Prof. Martin Barzyck, Dr. Gaurav Sinha, Dr. Ranjan Mondal, and Dr. Sanchayan Santra at various stages of the research journey.
Background
ML Researcher, currently researching efficient use of LLMs and evaluation of LLM-based Agents. Interested in sequential decision-making problems (bandits, RL, etc.). In the future, aims to contribute to AI for Science problems.
Miscellany
Links to LinkedIn, GitHub, CV, Google Scholar, and personal email provided.