Selected publications
We publish and open-source our work — papers, benchmarks, and tools. An active and reputable presence in academia is how we access top-tier talent, partner with model vendors, and keep a cross-product perspective.
Developer Needs and Feasible Features for AI Assistants in IDEs
Asks developers what they actually want from in-IDE assistance, then sorts those wants by what is feasible to build.
Human-AI Experience in Integrated Development Environments: A Systematic Literature Review
Maps what is known about human–AI interaction inside the IDE, and where the field's evidence actually stops.
The Complexity Trap: Simple Observation Masking Is as Efficient as LLM Summarization for Agent Context Management
A study of context-management strategies for coding agents — context compression and observation masking over long horizons.
AI in Software Engineering: Perceived Roles and Their Impact on Adoption
Developers cast AI tools either as an inanimate instrument or as a human-like teammate — and the more roles they assign one, the more useful they find it.
Drawing Pandas: A Benchmark for LLMs in Generating Plotting Code
175 human-curated data-analysis tasks that score plotting code by the picture it draws, judged by a vision model against the ground-truth plot.
Our record does not stop at this page.Google Scholar holds the wider body of work and citations.