About
I am an AI researcher and engineer based in Bengaluru, India. My research covers efficient fine-tuning (LoRA, QLoRA), federated learning, model compression, and how well language models reason after compression or across languages. At Synaptics I design and ship enterprise AI automation, including an invoice automation platform that is live for the Korea division. I studied for an M.Tech in Data Science at Dr. B.R. Ambedkar National Institute of Technology Jalandhar (2024–2026).
Research interests
- Efficient machine learning
- Parameter-efficient fine-tuning (LoRA, QLoRA)
- Federated learning
- Model compression and quantization
- Reasoning evaluation of language models
- Cross-lingual NLP
- Diffusion language models
Education
2024–2026
M.Tech in Data Science
Dr. B.R. Ambedkar National Institute of Technology Jalandhar
CGPA 7.80/10
Dissertation: Fine-Tuning Large Language Models for Indian Statutory Legal NLP using Low-Rank Adaptation
Supervisor: Dr. Amritpal Singh
2019–2023
B.Tech in Computer Science and Engineering
Shri Mata Vaishno Devi University (SMVDU)
CGPA 7.27/10
Languages
English, Hindi
Skills
- LLM and SLM
- PyTorch, Hugging Face (Transformers, PEFT, bitsandbytes, Datasets), LoRA, QLoRA, Federated learning
- Compression
- Quantization (NF4, AWQ, GPTQ, SmoothQuant, SparseGPT, Wanda, QAT), Pruning
- Theory and methods
- SVD and subspace tracking, Non-convex optimization, Convergence analysis, Statistical hypothesis testing (McNemar, Wilcoxon, Cohen's d)
- Engineering and cloud
- Python, FastAPI, Docker, Oracle SQL, TensorFlow, scikit-learn, Azure, OCI, NVIDIA H100 HPC