Work Experience
Capital OneApplied Researcher — AI Foundations
- AI Foundations team, Cambridge, Massachusetts, USA
- Research on efficient and production-ready Large Language Model training, post-training and inference
Red Hat Inc.Senior Machine Learning Research Engineer
- Drove research on post-training and inference, including mixed-precision quantization, parallel token drafting for speculative decoding, RL rollouts, and Agentic AI evaluation (e.g., SWE-Bench)
- Contributed to open-source frameworks: vLLM, llm-compressor, speculators and the Red Hat AI Hugging Face model repository, enabling production deployment of compressed and accelerated LLMs
- Integrated FP8 quantized Inkling, Qwen 3.6, Granite4 and Nemotron family of models into vLLM and llm-compressor; developed Dflash drafters (parallel token predictors) for Qwen3 and Nemotron
- Worked on tool calling and Agentic datasets and evaluation
- Mentored one intern on the KV-Dflash project
Argonne National LaboratoryPostdoctoral Researcher
- AI/ML team of Argonne's Leadership Computing Facility (ALCF)
- Research at the intersection of Systems and Deep Learning; optimizing inference and finetuning of LLMs
- Collaborated with scientists and engineers from Nvidia, Intel, AMD, SambaNova, Cerebras, Groq, Graphcore and Habana
Argonne National LaboratoryResearch Intern
- AI/ML team of Argonne's Supercomputing facility
- Worked on a project at the intersection of Pruning, Quantization and Neural Architecture Search
- Published research findings at HPDC 2022
Intel CorporationDeep Learning Research Scientist Intern
- Graphics Processing Research Lab
- Designed CNN model optimization strategies for Image Super Resolution and Denoising
- Designed NAS methods for Mixed Precision Quantization and hardware-aware neural networks
- Published research at ICIP 2021
Advanced Micro Devices (AMD)Deep Learning Intern — MIGraphX Team
- Developed Post Training Quantization (PTQ) methods to reduce CNN weights from FP32 to Int8
- Implemented quantization on VGG16, ResNet50, InceptionV3, Xception with negligible accuracy loss on ImageNet
Research Centre Imarat, DRDO
Undergraduate Technical Intern
Bharat Dynamics Limited (BDL)
Undergraduate Technical Intern