Tune GitHub Copilot Inline Suggestions with Configurable Eagerness
💡 How proactively should Copilot suggest code? You decide!
About me
I am a Senior Researcher at Microsoft Code|AI, based in Des Moines, Iowa.
I work on GitHub Copilot's code edit suggestion models and agents.
Posts & Projects
See all- Auto-Syncing Upstream Periodically to a Fork for Open-Source ContributionJul 2026 · Blog
- Windows Terminal profile for Claude with no permission promptsMar 2026 · Blog
- Tailscale + SSH setup for Windows + WSL2Mar 2026 · Blog
- wslwatch: Watchdog that keeps WSL2 distributions runningMar 2026 · Project
- rrun: Git-aware remote command runner over SSHMar 2026 · Project
- How to limit the number of Twitter posts in your timeline using JavaScriptMar 2024 · Blog
- How to set up a shared cache for HuggingFace librariesJan 2024 · Blog
- How to fix
docker-compose: command not founderror with newer versions of DockerDec 2023 · Note - How to use task-spooler as a shared queueing systemDec 2023 · Blog
- shared-task-spooler: Convenience script for managing a shared job queue with task-spoolerDec 2023 · Project
- friendship ended with
earlyoom, nownohangis my best friendFeb 2023 · Note - Siamese network and triplet lossOct 2022 · Note
- Get filepath of Bash activation scriptJul 2022 · Note
- task-spooler-gui: Web interface for monitoring and accessing task-spooler queuesJul 2022 · Project
- webcam-mods for Linux background blur & swapJul 2022 · Note
- Beware any vs lenJul 2022 · Note
- Use head -n -0 to get all items in listJul 2022 · Note
- Use hard links to replicate log files in a generated directoryJun 2022 · Note
- Beware metric auto-reduce with PyTorch Lightning + TorchMetricsJun 2022 · Note
- Print pandas Series as percentJun 2022 · Note
- Encoder-decoder modelsJun 2022 · Note
- Print item keys with jqJun 2022 · Note
- Neural Network EmbeddingsJun 2022 · Note
- tree-climber: Program analysis tools for C built on tree-sitterJun 2022 · Project
- Meditations on Meditations: I-IVJun 2022 · Blog
- Programming problem: GematriaMay 2022 · Blog
- No pain no gain? Comparing 3 program analysis frameworks for CMar 2022 · Blog
- cfactor: Policy-driven refactoring for C programsJul 2021 · Project
- animal-cognitive: Embodied-cognition reinforcement learning modelsMar 2021 · Project
- pal-tools: Dynamic analysis and code-generation utilitiesDec 2020 · Project
- rarl: Reproduction of "robust adversarial reinforcement learning"Nov 2020 · Project
- precise-interrupts: ARM interrupt-handling research reproductionOct 2020 · Project
Selected publications
See all on Scholar2026
CodeSense: a Real-World Benchmark and Dataset for Code Semantic Reasoning
We propose CodeSense, the first benchmark with a spectrum of fine-grained code reasoning tasks from real-world software projects, revealing a clear performance gap in state-of-the-art LLMs for code semantic reasoning.
Saving SWE-Bench: A Benchmark Mutation Approach for Realistic Agent Evaluation
We introduce a benchmark mutation framework that transforms formal GitHub issue benchmarks into realistic user-style queries, revealing that existing benchmarks overestimate agent capabilities by up to 50%.
2025
Closing the Gap: A User Study on the Real-world Usefulness of AI-powered Vulnerability Detection & Repair in the IDE
We present DeepVulGuard, an AI-powered vulnerability detection & repair tool built into the VSCode IDE, and the results of a user study on this tool with 17 professional software developers.
Reinforcement Learning from Automatic Feedback for High-Quality Unit Test Generation
We propose a novel technique called Reinforcement Learning from Static Quality Metrics (RLSQM).
2024
Dataflow Analysis-Inspired Deep Learning for Efficient Vulnerability Detection
We present DeepDFA, a dataflow analysis-inspired graph learning framework and an embedding technique that enables graph learning to simulate dataflow computation.
TRACED: Execution-aware Pre-training for Source Code
We introduce TRACED, an execution-aware pre-training strategy for source code wherein we pre-train code language models with a combination of source code, executable inputs, and corresponding execution traces.
2023
An Empirical Study of Deep Learning Models for Vulnerability Detection
In this paper, we surveyed and reproduced 9 state-of-the-art (SOTA) deep learning models on 2 widely used vulnerability detection datasets: Devign and MSR.