Enhanced performance metrics for quantized LLMs by expanding support from 3 to 22+ quantization methods (including GPTQ, AWQ, BitsAndBytes, and FP8 variants). Refactored quantization parsing logic and introduced unified weight mappings to improve accuracy of FLOPs and memory estimation, with added test coverage for robustness.

Focusing on LLM knowledge editing, investigating methods to update and refine model knowledge without full retraining.

Designing evaluation pipelines, running large-scale experiments, and analyzing robustness, locality, and generalization of edits.

Developed an autonomous LLM-powered web crawler to extract and validate KYC documents from multiple regulatory websites.

Implemented an agentic RAG architecture with scalable storage and optimized retrieval for efficient access to compliance information.

Conducted research on X-ray radiography images from welding inspections, developing deep learning models to detect fusion flaws with high precision.

Worked alongside engineers to streamline inspection workflows using MLOps pipelines and enhance the reliability of quality checks.

Researched advanced transformer architectures for medical imaging, implementing UniFormer to analyze multi-phase MRI scans for liver lesion diagnosis.

Applied specialized loss functions to address class imbalance and ranked among the top 15 teams globally in the MICCAI Liver Lesion Challenge.