AI Systems · GPU Infrastructure · Agentic AI on HPC
Aditya Chopra
Computer Science student at Stevens Institute of Technology building systems at the intersection of LLM inference, CUDA kernels, distributed GPU computing, and quant infrastructure.
Agentic AI on Polaris
Deploying local LLM and agentic workflows on ALCF Polaris using vLLM, OpenAI-compatible APIs, Ray, tensor parallelism, and A100 GPUs.
FlashAttention + Paged Attention
CUDA attention kernels comparing vanilla attention, FlashAttention-style prefill, and paged KV-cache decode.
RL Trading Agent
Reinforcement learning trading experiments using SHIFT Simulator, Stable-Baselines3, PPO, DQN, reward shaping, and inventory-aware observations.
Market Anomaly Detection
Quant ML project involving feature engineering, model comparison, class imbalance handling, and precision-recall evaluation.