Yaroslav Golubev

Happy to announce the first keynote at the @IDEworkshop co-located with @ICSEconf! "Does JetBrains Dream of HAX? Human-AI Experience in the IDE, Today and Tomorrow" by Agnia Sergeyuk, Team Lead of Human-AI Experience Research at @jetbrains! 👏Join us in Rio!

We accepted 11 great and diverse papers! Come join us in Rio, and let's talk IDEs!

The Best of N Worlds Aligning Reinforcement Learning with Best-of-N Sampling via max@k Optimisation

⏰ The deadline for submissions is extended to October 27th! 🥳🥳🥳 One more week to submit your works! ⏰

Our Context Collection Competition co-organized by JetBrains and Mistral AI wrapped up! 🎉 Congratulations to the winners in both Python and Kotlin code completion tracks! We just published a report that describes the competition and the key approaches: arxiv.org/abs/2510.04349

🥳 We were happy to notice that the papers published in the first two editions of the IDE Workshop collectively passed 100 citations! 📜 You can find all of them with pre-prints here: scholar.google.com/citations?user… ⏰ Please remember to submit to the 3rd IDE Workshop by October 20th!

One of our core values is 🌉 bridging academia and industry 🌉. We are happy to share that two of our papers were just accepted to ASE'25 Industry, both from the AI4SE collaboration with @tudelft. 🪶 TreeRanker: Fast and Model-agnostic Ranking System for Code Suggestions in IDEs 🔗 arxiv.org/abs/2508.02455 ☝ A new scoring approach to ranking static completions using language models in a lightweight and model-agnostic way, employing a prefix tree. 🤝 Congratulations to @dan_cip00, @egor_bb, @avandeursen, and @MalihehIzadi! 🥳 🪶 Prompt-with-Me: in-IDE Structured Prompt Management for LLM-Driven Software Engineering 🔗 arxiv.org/abs/2509.17096 ☝ A system for prompt management that automatically classifies prompts using a four-dimensional taxonomy, suggests language refinements, masks sensitive information, and extracts reusable templates from a developer’s prompt library. 🤝 Congratulations to Ziyou, Agnia, and @MalihehIzadi! 🥳

The @IDEWorkshop is a research conference about everything IDEs, co-organized by the folks at JetBrains Research! 🚀 The next one will take place at @ICSEconf in Brazil in April 2026. Submissions from industry are also welcome! Full info: jb.gg/ide-workshop-2…

📢 At @IEEEICSME, @danny_dig just presented our paper "Together We Are Better: LLM, IDE and Semantic Embedding to Assist Move Method Refactoring"! Check out the demo-video and the tool! 💛 📜: fraolbatole.github.io/assets/pdf/MMA… 🛠️: mm-assist.netlify.app/installation 🎬: #demo-video" target="_blank" rel="nofollow noopener">mm-assist.netlify.app/#demo-video

The deadline for submissions is October 20th! We are eager to receive your works on absolutely anything related to IDEs! If you built new plugins, evaluated something on IDE users, or in general have brand new ideas about how IDEs should be organized — join us! ❤️

We hope KOALA can be useful to you for your IDE-based experiments in SE research! 📜 Paper: arxiv.org/abs/2506.21266 🛠️ Tool: github.com/JetBrains-Rese… 💾 Dataset: zenodo.org/records/155533…

Another paper at CompEd is 🐨 "KOALA: a Configurable Tool for Collecting IDE Data When Solving Programming Tasks"!🐨 The IDE plugin can provide the students with the necessary tasks, enable or disable IDE features, and collect code snapshots and all the IDE activity!

🎓 The largest survey paper I ever participated in was accepted to CompEd! Opinions of ~18k learners from 173 countries about all the aspects of CS education (87 questions!) — in the form of open dataset. Enjoy! 💡 Paper: arxiv.org/abs/2508.05286 Dataset: zenodo.org/records/167541…

"AI in Software Engineering: Perceived Roles and Their Impact on Adoption", the AIIDE workshop @FSEconf. 🤖 Diving deeper into the same topic, we identified two primary "mental models" of AI in developers: 🛠️ AI as an inanimate tool and 👨🏻‍💼 AI as a human-like teammate. We showed that assigning multiple roles to AI correlates positively with Perceived Usefulness and Perceived Ease of Use for AI tools, indicating that diverse conceptualizations enhance AI adoption. These insights suggest that AI4SE tools should accommodate varying user expectations through adaptive design strategies that align with different mental models. 📜 Paper: arxiv.org/abs/2504.20329

"From Teacher to Colleague: How Coding Experience Shapes Developer Perceptions of AI Tools", the Generative AI and HCI workshop at @acm_chi. 🤝 The paper is based on the JetBrains Developer Ecosystem Report 2024, and explores how coding experience affects AI awareness, AI adoption, and the roles developers assign to AI in their workflow. Key takeaways: 💡 Coding experience doesn’t predict AI adoption — but it does shape how people think about AI’s role. 💡 More experienced devs tend to see AI as a junior colleague, content generator, or even irrelevant. 💡 Less experienced devs more often view AI as a teacher. 📜 Paper: arxiv.org/abs/2504.13903

🤔 How do people perceive AI? 🤔 In two recent publications, we investigated how developers perceive AI tools, how it affects their satisfaction with these tools and the adoption of new ones. See below!🧵

Happy to launch the third edition of the @IDEworkshop co-located with @ICSEconf 2026! Please follow the links, submit your papers on anything IDE-related, and let's meet in Brazil to continue building this amazing community! 🎉

📢 In just several minutes, our folks will be presenting their two papers at the DEBT workshop co-located with ECOOP'25! 📊 "Automatic Data Structures Visualization in IntelliJ IDEA" 🐞 "Trace Debugger: Interactive Execution Trace Debugging for Java and Kotlin" Catch them!

Two of our talks from the EuroProofNet WG5 meeting! 🕵 🔎 "Are LLMs Ready for Software Verification?". We create benchmarks for Nagini and Verus based on HumanEval and test Claude Sonnet 3.5 in software verification. 📜 Abstract: europroofnet.github.io/_pages/WG5/Edi… 💾 Data: github.com/JetBrains-Rese… 🐓 "CoqPilot Benchmarking Framework". We present CoqPilot, a benchmark for generating Coq proofs using LLMs. 📜 Abstract: europroofnet.github.io/_pages/WG5/Edi… 💾 Data: github.com/JetBrains-Rese… Congratulations to @inf_anlun and everyone involved!

📣 Another one of our papers will be presented at @FSEconf next week: "A Multi-agent Onboarding Assistant based on Large Language Models, Retrieval Augmented Generation, and Chain-of-Thought" 🤩 This work is about an 🪵 onboarding agent 🕵🏻 designed to assist new team members in understanding a codebase. The agent helps answer questions about the repository using a combination of techniques, including Retrieval-Augmented Generation (RAG) and Chain-of-Thought (CoT) prompting. Most importantly, we conducted a 👤 human study 🎓 to evaluate how helpful the agent is during the onboarding process. Participants rated the agent as helpful and useful in navigating and understanding the codebase. We believe that the rise of LLMs will not only enhance code generation and issue resolution but also significantly improve how people explore and comprehend large codebases! 📜 Paper: arxiv.org/abs/2503.23421 🛠️ Tool: onboarding.software/register?colab… 🎬 Video: youtube.com/watch?v=V18mjp… Kudos to Andrei-Cristian, @smtitov, and @MalihehIzadi!