arXiv subjects

Yaroslav Golubev

Publications and source records attributed to Yaroslav Golubev.

At least 19 recordsLinked to original sources

arXiv2026

Observing Fine-Grained Changes in Jupyter Notebooks During Development Time

In software engineering research, the analysis of fine-grained logs led to significant innovations in areas such as refactoring, security, and code completion. However, even though computational notebooks are a staple of data science and an important tool in machine learning, few similar studies have been conducted in this area. To help bridge this research gap, this paper makes three scientific contributions. (1) We introduce a toolset for collecting code changes in Jupyter notebooks during development time. (2) We use it to collect more than 100 hours of work related to a data analysis task and a machine learning task (carried out by 20 developers with different levels of expertise), resulting in a dataset containing 2,655 cells and 9,207 cell executions. (3) Finally, we use this dataset to investigate the dynamic nature of the notebook development process and the changes that take place in the notebooks. In our analysis of the collected data, we classified the changes made to the cells between executions and found that a significant number of these changes constituted code iteration modifications. We report a number of other insights and propose detailed future research directions on the novel data.

cs.SE↗

arXiv2026

Evolving with AI: A Longitudinal Analysis of Developer Logs

AI-powered coding assistants are rapidly becoming fixtures in professional IDEs, yet their sustained influence on everyday development remains poorly understood. Prior research has focused on short-term use or self-reported perceptions, leaving open questions about how sustained AI use reshapes actual daily coding practices in the long term. We address this gap with a mixed-method study of AI adoption in IDEs, combining longitudinal two-year fine-grained telemetry from 800 developers with a survey of 62 professionals. We analyze five dimensions of workflow change: productivity, code quality, code editing, code reuse, and context switching. Telemetry reveals that AI users produce substantially more code but also delete significantly more. Meanwhile, survey respondents report productivity gains and perceive minimal changes in other dimensions. Our results offer empirical insights into the silent restructuring of software workflows and provide implications for designing future AI-augmented tooling.

cs.SE↗

arXiv2025

Diff-XYZ: A Benchmark for Evaluating Diff Understanding

Reliable handling of code diffs is central to agents that edit and refactor repositories at scale. We introduce Diff-XYZ, a compact benchmark for code-diff understanding with three supervised tasks: apply (old code $+$ diff $\rightarrow$ new code), anti-apply (new code $-$ diff $\rightarrow$ old code), and diff generation (new code $-$ old code $\rightarrow$ diff). Instances in the benchmark are triples $\langle \textit{old code}, \textit{new code}, \textit{diff} \rangle$ drawn from real commits in CommitPackFT, paired with automatic metrics and a clear evaluation protocol. We use the benchmark to do a focused empirical study of the unified diff format and run a cross-format comparison of different diff representations. Our findings reveal that different formats should be used depending on the use case and model size. For example, representing diffs in search-replace format performs best for larger models across most tasks, while structured udiff variants offer similar but slightly weaker performance. In contrast, smaller open models benefit little from any formatting choice. The Diff-XYZ benchmark is a reusable foundation for assessing and improving diff handling in LLMs that can aid future development of diff formats and models editing code. The dataset is published on HuggingFace Hub: https://huggingface.co/datasets/JetBrains-Research/diff-xyz.

cs.SE↗

arXiv2025

Understanding Student Interaction with AI-Powered Next-Step Hints: Strategies and Challenges

Automated feedback generation plays a crucial role in enhancing personalized learning experiences in computer science education. Among different types of feedback, next-step hint feedback is particularly important, as it provides students with actionable steps to progress towards solving programming tasks. This study investigates how students interact with an AI-driven next-step hint system in an in-IDE learning environment. We gathered and analyzed a dataset from 34 students solving Kotlin tasks, containing detailed hint interaction logs. We applied process mining techniques and identified 16 common interaction scenarios. Semi-structured interviews with 6 students revealed strategies for managing unhelpful hints, such as adapting partial hints or modifying code to generate variations of the same hint. These findings, combined with our publicly available dataset, offer valuable opportunities for future research and provide key insights into student behavior, helping improve hint design for enhanced learning support.

cs.SE↗

arXiv2025

Mellum: Production-Grade in-IDE Contextual Code Completion with Multi-File Project Understanding

We present the Mellum models family, open-weight code completion models designed for interactive use in JetBrains IDEs. Mellums have 4B parameters, adopt a Llama-style architecture, and are pre-trained on ~4T tokens of permissively licensed, multi-language code. Our studies show that (i) careful data curation and staged training significantly improve the model's quality, (ii) editor-critical capabilities such as context packing are necessary for high-quality suggestions, and (iii) a compact, task-focused model can meet the cost and latency constraints of interactive completion. In the paper, we describe an end-to-end industrial pipeline for producing contextualized in-editor completion: disciplined data governance, multi-stage training that includes fill-in-the-middle and project context via supervised fine-tuning, and alignment via direct preference optimization using feedback from real-world scenarios. Our quality evaluations include both large-scale offline benchmarks and online telemetry from production deployments in JetBrains IDEs. Mellums are released under the Apache-2.0 license on HuggingFace, with a public model card providing a reproducible reference for practitioners. Our experience offers a pragmatic blueprint for taking a focused, open model from a research prototype to at scale production for hundreds of thousands of users.

cs.SE↗

arXiv2025

Challenge on Optimization of Context Collection for Code Completion

The rapid advancement of workflows and methods for software engineering using AI emphasizes the need for a systematic evaluation and analysis of their ability to leverage information from entire projects, particularly in large code bases. In this challenge on optimization of context collection for code completion, organized by JetBrains in collaboration with Mistral AI as part of the ASE 2025 conference, participants developed efficient mechanisms for collecting context from source code repositories to improve fill-in-the-middle code completions for Python and Kotlin. We constructed a large dataset of real-world code in these two programming languages using permissively licensed open-source projects. The submissions were evaluated based on their ability to maximize completion quality for multiple state-of-the-art neural models using the chrF metric. During the public phase of the competition, nineteen teams submitted solutions to the Python track and eight teams submitted solutions to the Kotlin track. In the private phase, six teams competed, of which five submitted papers to the workshop.

cs.SE↗

arXiv2025

Everything You Need to Know About CS Education: Open Results from a Survey of More Than 18,000 Participants

Computer science education is a dynamic field with many aspects that influence the learner's path. While these aspects are usually studied in depth separately, it is also important to carry out broader large-scale studies that touch on many topics, because they allow us to put different results into each other's perspective. Past large-scale surveys have provided valuable insights, however, the emergence of new trends (e.g., AI), new learning formats (e.g., in-IDE learning), and the increasing learner diversity highlight the need for an updated comprehensive study. To address this, we conducted a survey with 18,032 learners from 173 countries, ensuring diverse representation and exploring a wide range of topics - formal education, learning formats, AI usage, challenges, motivation, and more. This paper introduces the results of this survey as an open dataset, describes our methodology and the survey questions, and highlights, as a motivating example, three possible research directions within this data: challenges in learning, emerging formats, and insights into the in-IDE format. The dataset aims to support further research and foster advancements in computer education.

cs.CY↗

arXiv2025

KOALA: a Configurable Tool for Collecting IDE Data When Solving Programming Tasks

Collecting data of students solving programming tasks is incredibly valuable for researchers and educators. It allows verifying that the students correctly apply the features and concepts they are taught, or finding students' misconceptions. However, existing data collection tools have limitations, e.g., no control over the granularity of the collected code, not collecting the specific events of the programming environment used, and overall being hard to configure. To overcome these limitations, we propose KOALA, a convenient and highly configurable tool for collecting code snapshots and feature usage from students solving programming tasks in JetBrains IDEs. The plugin can be installed in IDEs and configured to provide the students with the necessary tasks, enable or disable certain IDE features like code completion, and run surveys. During problem solving, the plugin collects code snapshots at the configured granularity, all IDE actions like running and debugging, as well as some data not collected in prior works, like employed hotkeys and switching focus between files. The collected data is sent to the server that comes with the tool, where it is stored and can be converted to the standardized ProgSnap2 format. To showcase the tool, we collected data from 28 students solving tasks in two courses within the IDE, highlighting some insights from this data.

cs.SE↗

arXiv2025

LitmusKt: Concurrency Stress Testing for Kotlin

We present LitmusKt - the first tool for litmus testing concurrent programs in Kotlin. The tool's novelty also lies in the fact that Kotlin is a multiplatform language, i.e., it compiles into multiple platforms, which means that the concurrency has to be tested on several of them. Our tool allows writing litmus tests in a single custom DSL, and these tests are then run in Kotlin/Native and Kotlin/JVM, two main platforms for concurrent programming in Kotlin. Using LitmusKt, we discovered novel bugs in the Kotlin compiler, which we then fixed and they are no longer present. Moreover, LitmusKt was integrated into the CI pipeline for Kotlin. LitmusKt is available on GitHub: https://github.com/JetBrains-Research/litmuskt. The demo is available on YouTube: https://youtu.be/oWCZp_Huwss.

cs.PL↗

arXiv2025

Drawing Pandas: A Benchmark for LLMs in Generating Plotting Code

This paper introduces the human-curated PandasPlotBench dataset, designed to evaluate language models' effectiveness as assistants in visual data exploration. Our benchmark focuses on generating code for visualizing tabular data - such as a Pandas DataFrame - based on natural language instructions, complementing current evaluation tools and expanding their scope. The dataset includes 175 unique tasks. Our experiments assess several leading Large Language Models (LLMs) across three visualization libraries: Matplotlib, Seaborn, and Plotly. We show that the shortening of tasks has a minimal effect on plotting capabilities, allowing for the user interface that accommodates concise user input without sacrificing functionality or accuracy. Another of our findings reveals that while LLMs perform well with popular libraries like Matplotlib and Seaborn, challenges persist with Plotly, highlighting areas for improvement. We hope that the modular design of our benchmark will broaden the current studies on generating visualizations. Our dataset and benchmark code are available online: https://huggingface.co/datasets/JetBrains-Research/PandasPlotBench; https://github.com/JetBrains-Research/PandasPlotBench.

cs.SE↗

arXiv2025

Finding Important Stack Frames in Large Systems

In this work, we developed, integrated, and tested a feature that automatically highlights potentially important frames in stack traces. The feature was implemented in the internal bug-processing tool at JetBrains that processes tens of millions of stack traces. We surveyed 18 developers at JetBrains who provided valuable feedback on the idea and the implementation.

cs.SE↗

arXiv2025

Full Line Code Completion: Bringing AI to Desktop

In recent years, several industrial solutions for the problem of multi-token code completion appeared, each making a great advance in the area but mostly focusing on cloud-based runtime and avoiding working on the end user's device. In this work, we describe our approach for building a multi-token code completion feature for the JetBrains' IntelliJ Platform, which we call Full Line Code Completion. The feature suggests only syntactically correct code and works fully locally, i.e., data querying and the generation of suggestions happens on the end user's machine. We share important time and memory-consumption restrictions, as well as design principles that a code completion engine should satisfy. Working entirely on the end user's device, our code completion engine enriches user experience while being not only fast and compact but also secure. We share a number of useful techniques to meet the stated development constraints and also describe offline and online evaluation pipelines that allowed us to make better decisions. Our online evaluation shows that the usage of the tool leads to 1.3 times more Python code in the IDE being produced by code completion. The described solution was initially started with a help of researchers and was then bundled into all JetBrains IDEs where it is now used by millions of users. Thus, we believe that this work is useful for bridging academia and industry, providing researchers with the knowledge of what happens when complex research-based solutions are integrated into real products.

cs.SE↗

arXiv2025

Towards Realistic Evaluation of Commit Message Generation by Matching Online and Offline Settings

When a Commit Message Generation (CMG) system is integrated into the IDEs and other products at JetBrains, we perform online evaluation based on user acceptance of the generated messages. However, performing online experiments with every change to a CMG system is troublesome, as each iteration affects users and requires time to collect enough statistics. On the other hand, offline evaluation, a prevalent approach in the research literature, facilitates fast experiments but employs automatic metrics that are not guaranteed to represent the preferences of real users. In this work, we describe a novel way we employed to deal with this problem at JetBrains, by leveraging an online metric - the number of edits users introduce before committing the generated messages to the VCS - to select metrics for offline experiments. To support this new type of evaluation, we develop a novel markup collection tool mimicking the real workflow with a CMG system, collect a dataset with 57 pairs consisting of commit messages generated by GPT-4 and their counterparts edited by human experts, and design and verify a way to synthetically extend such a dataset. Then, we use the final dataset of 656 pairs to study how the widely used similarity metrics correlate with the online metric reflecting the real users' experience. Our results indicate that edit distance exhibits the highest correlation with the online metric, whereas commonly used similarity metrics such as BLEU and METEOR demonstrate low correlation. This contradicts the previous studies on similarity metrics for CMG, suggesting that user interactions with a CMG system in real-world settings differ significantly from the responses by human labelers within controlled environments. We release all the code and the dataset to support future research in the field: https://jb.gg/cmg-evaluation.

cs.SE↗

arXiv2024

Stack Trace Deduplication: Faster, More Accurately, and in More Realistic Scenarios

In large-scale software systems, there are often no fully-fledged bug reports with human-written descriptions when an error occurs. In this case, developers rely on stack traces, i.e., series of function calls that led to the error. Since there can be tens and hundreds of thousands of them describing the same issue from different users, automatic deduplication into categories is necessary to allow for processing. Recent works have proposed powerful deep learning-based approaches for this, but they are evaluated and compared in isolation from real-life workflows, and it is not clear whether they will actually work well at scale. To overcome this gap, this work presents three main contributions: a novel model, an industry-based dataset, and a multi-faceted evaluation. Our model consists of two parts - (1) an embedding model with byte-pair encoding and approximate nearest neighbor search to quickly find the most relevant stack traces to the incoming one, and (2) a reranker that re-ranks the most fitting stack traces, taking into account the repeated frames between them. To complement the existing datasets collected from open-source projects, we share with the community SlowOps - a dataset of stack traces from IntelliJ-based products developed by JetBrains, which has an order of magnitude more stack traces per category. Finally, we carry out an evaluation that strives to be realistic: measuring not only the accuracy of categorization, but also the operation time and the ability to create new categories. The evaluation shows that our model strikes a good balance - it outperforms other models on both open-source datasets and SlowOps, while also being faster on time than most. We release all of our code and data, and hope that our work can pave the way to further practice-oriented research in the area.

cs.SE↗

arXiv2024

Using AI-Based Coding Assistants in Practice: State of Affairs, Perceptions, and Ways Forward

Context. The last several years saw the emergence of AI assistants for code - multi-purpose AI-based helpers in software engineering. As they become omnipresent in all aspects of software development, it becomes critical to understand their usage patterns. Objective. We aim to better understand how specifically developers are using AI assistants, why they are not using them in certain parts of their development workflow, and what needs to be improved in the future. Methods. In this work, we carried out a large-scale survey aimed at how AI assistants are used, focusing on specific software development activities and stages. We collected opinions of 481 programmers on five broad activities: (a) implementing new features, (b) writing tests, (c) bug triaging, (d) refactoring, and (e) writing natural-language artifacts, as well as their individual stages. Results. Our results provide a novel comparison of different stages where AI assistants are used that is both comprehensive and detailed. It highlights specific activities that developers find less enjoyable and want to delegate to an AI assistant, e.g., writing tests and natural-language artifacts. We also determine more granular stages where AI assistants are used, such as generating tests and generating docstrings, as well as less studied parts of the workflow, such as generating test data. Among the reasons for not using assistants, there are general aspects like trust and company policies, as well as more concrete issues like the lack of project-size context, which can be the focus of the future research. Conclusion. The provided analysis highlights stages of software development that developers want to delegate and that are already popular for using AI assistants, which can be a good focus for features aimed to help developers right now. The main reasons for not using AI assistants can serve as a guideline for future work.

cs.SE↗

arXiv2024

Clustering MOOC Programming Solutions to Diversify Their Presentation to Students

In many MOOCs, whenever a student completes a programming task, they can see previous solutions of other students to find potentially different ways of solving the problem and to learn new coding constructs. However, a lot of MOOCs simply show the most recent solutions, disregarding their diversity or quality, and thus hindering the students' opportunity to learn. In this work, we explore this novel problem for the first time. To solve it, we adapted the existing plagiarism detection tool JPlag to Python submissions on Hyperskill, a popular MOOC platform. However, due to the tool's inner algorithm, JPLag fully processed only 46 out of 867 studied tasks. Therefore, we developed our own tool called Rhubarb. This tool first standardizes solutions that are algorithmically the same, then calculates the structure-aware edit distance between them, and then applies clustering. Finally, it selects one example from each of the largest clusters, thus ensuring their diversity. Rhubarb was able to handle all 867 tasks successfully. We compared different approaches on a set of 59 real-life tasks that both tools could process. Eight experts rated the selected solutions based on diversity, code quality, and usefulness. The default platform approach of simply selecting recent submissions received on average 3.12 out of 5, JPlag - 3.77, Rhubarb - 3.50. To ensure both quality and coverage, we created a system that combines both tools. We conclude our work by discussing the future of this new problem and the research needed to solve it better.

cs.SE↗

arXiv2024

One Step at a Time: Combining LLMs and Static Analysis to Generate Next-Step Hints for Programming Tasks

Students often struggle with solving programming problems when learning to code, especially when they have to do it online, with one of the most common disadvantages of working online being the lack of personalized help. This help can be provided as next-step hint generation, i.e., showing a student what specific small step they need to do next to get to the correct solution. There are many ways to generate such hints, with large language models (LLMs) being among the most actively studied right now. While LLMs constitute a promising technology for providing personalized help, combining them with other techniques, such as static analysis, can significantly improve the output quality. In this work, we utilize this idea and propose a novel system to provide both textual and code hints for programming tasks. The pipeline of the proposed approach uses a chain-of-thought prompting technique and consists of three distinct steps: (1) generating subgoals - a list of actions to proceed with the task from the current student's solution, (2) generating the code to achieve the next subgoal, and (3) generating the text to describe this needed action. During the second step, we apply static analysis to the generated code to control its size and quality. The tool is implemented as a modification to the open-source JetBrains Academy plugin, supporting students in their in-IDE courses. To evaluate our approach, we propose a list of criteria for all steps in our pipeline and conduct two rounds of expert validation. Finally, we evaluate the next-step hints in a classroom with 14 students from two universities. Our results show that both forms of the hints - textual and code - were helpful for the students, and the proposed system helped them to proceed with the coding tasks.

cs.SE↗

arXiv2024

Long Code Arena: a Set of Benchmarks for Long-Context Code Models

Nowadays, the fields of code and natural language processing are evolving rapidly. In particular, models become better at processing long context windows - supported context sizes have increased by orders of magnitude over the last few years. However, there is a shortage of benchmarks for code processing that go beyond a single file of context, while the most popular ones are limited to a single method. With this work, we aim to close this gap by introducing Long Code Arena, a suite of six benchmarks for code processing tasks that require project-wide context. These tasks cover different aspects of code processing: library-based code generation, CI builds repair, project-level code completion, commit message generation, bug localization, and module summarization. For each task, we provide a manually verified dataset for testing, an evaluation suite, and open-source baseline solutions based on popular LLMs to showcase the usage of the dataset and to simplify adoption by other researchers. We publish the benchmark page on HuggingFace Spaces with the leaderboard, links to HuggingFace Hub for all the datasets, and link to the GitHub repository with baselines: https://huggingface.co/spaces/JetBrains-Research/long-code-arena.

cs.LG↗