Hello, I'm Arpit Singh Gautam

I am a Data Scientist in the CSG CTO Lab at Dell Technologies, working on efficient LLM inference and the reliability of large models. My research identity is Efficient × Trustworthy Foundation Models: making large models cheaper to run, and understanding what compression does to their reliability and safety.

My deepest current line is the reliability of quantized models - what post-training quantization does to calibration, factual recall, and security, and how to preserve each cheaply. My work appears at EACL 2026 (FEVER) and AAAI 2026 (ToM), with PRISM accepted at Discover Computing and a preprint on RL-based quantization (RAMP).

Research interests: LLM systems & efficient inference - quantization, KV-cache optimization, serving · Trustworthy / honest LLMs - hallucination, calibration, safety · Reinforcement learning & reasoning for foundation models · Interpretability / mechanistic ML