Recent Work
Writing & Papers
- Attack Selection in Agentic AI Control Evaluations Meaningfully Decreases Safety opens in a new tab
Paper, ICML 2026 AI4GOOD, AI-WILD, and FAGEN, Jun 2026
- When Offline Selectors Cannot Beat the Best Single Model: A Diagnostic Study on edX Dropout Prediction opens in a new tab
Paper, ICML 2026 DEMO Workshop, Jun 2026
- Same Facts, Different Updates: Inference Setup Shapes LLM Behavior in Medical Allocation opens in a new tab
Paper, ICML 2026 AI4GOOD + Pluralistic Alignment Workshops, May 2026
- Attack Selection in Agentic AI Control Evals Can Decrease Safety opens in a new tab
Essay, LessWrong, Apr 2026
- Asymmetric Goal Drift in Coding Agents Under Value Conflict opens in a new tab
Paper, AI-WILD Workshop at ICLR 2026, Mar 2026
- Inherited Goal Drift: Contextual Pressure Can Undermine Agentic Goals opens in a new tab
Paper, AI-WILD Workshop at ICLR 2026, Mar 2026
Talks
- Hierarchical Reasoning Models
Talk, Latent Space Paper Club, Sep 2025
- Language Diffusion Survey
Talk, Latent Space Paper Club, May 2025
Selected Posts
all ideas →
A macOS menu bar app for Claude Code & Codex
Extending an MIT-licensed menu bar app to show Claude Code and OpenAI Codex usage side by side
Git Worktrees, branches without the context switch
Work on multiple branches at once without ever touching git stash.
CUDA Fundamentals: Tiled Matrix Multiply & Bitonic Sort
Writing real GPU kernels, exploring shared memory tiling, parallel sorting algorithms, and performance optimization on an NVIDIA H100.
Notes on Effective ML Research
A summary of my takeaways from three influential articles on conducting effective empirical AI alignment research.
Using uv: A Modern Python Workflow
An introduction to uv, a fast, Rust-based tool for Python packaging
Concurrency and Parallelism
A dive into concurrency and parallelism across different levels of computer systems
Master of Science in Computer Science
From biomedical engineering to machine learning: my journey through Georgia Tech's MSCS program
Language Diffusion Survey
A talk surveying diffusion models for language, from DDPM foundations to modern mask diffusion competing with auto-regressive models.