Projects

Selected research projects spanning methods, systems, experiments, and ongoing work.

Safe closed-loop agentic dose-response experiments using MHS

agents

lab-automation

mhs

During the limited Model Hardware Standard (MHS) research preview (announced by Anthropic), I led Carnegie Mellon’s work using MHS. The Model Hardware Standard is a new…

simPCL: An Open-Framework Simulator and Benchmark Generator for Programmable Cloud Laboratories

simulation

agents

benchmarking

ml-systems

Evaluating autonomous agents for experimental orchestration on real laboratory hardware is slow and expensive. Progress on planning, fault recovery, closed-loop scientific…

FlashAttention and PagedAttention in MiniTorch

ml-systems

llms

attention

Efficient attention and memory management are central to training and serving large language models. Coursework in LLM Systems at CMU focused on implementing these ideas in…

Agentic Orchestration and Fault Recovery

agents

rl-llm

decision-making

Autonomous systems that orchestrate multi-step workflows must plan actions, respond to failures, and adapt as conditions change. Fixed scripts are often too rigid for this…

Evolution-Informed Structural Subgraph Protein Representations

protein-engineering

graph-methods

foundation-models

Foundation models like ESM-2 provide powerful protein embeddings, while explicit structural and evolutionary features may add complementary information for function…

Reinforcement Learning for Senescence Biomarker Discovery

reinforcement-learning

biomarkers

single-cell

Identifying minimal gene panels that distinguish senescent cells from healthy cells is challenging due to high-dimensional data and complex cellular states in unlabeled data.