Find open source projects actively accepting contributors. Search repositories, filter by program milestones, difficulty tags, or tech stack.
Use our Orbit AI Matcher to find out! Get instant matching scores based on your developer skills, preferred frameworks, and contribution experience.
Convert your selected open-source project into a winning GSoC, LFX, or Outreachy application using Proposal Studio.
In the upcoming GSoC project, I will be compiling Stdlib codebase using LFortran. This codebase can be compiled by any modern Fortran compiler, such as GFortran. For various packages of Stdlib, my plan is to discover bugs and failures in various phases of LFortran (Frontend, ASR passes, and Backend) and report minimal reproducible examples for these bugs. My next step will be to fix all the reported bugs to ensure the successful generation of executables. Finally, I will be working to align the outputs from LFortran, for various Stdlib packages, with those of GFortran. Additionally, I will also work to implement some new features like Submodules which are required for the compilation of Stdlib but are not currently supported by LFortran.
To benchmark vector search algorithms (aka ANN algorithms), existing datasets fail to represent actual real-world workloads. Current benchmarks usually use small vectors with only a few hundred dimensions unlike modern embedding models like OpenAI's text-embedding series. This project aims to create a comprehensive dataset with 1 billion embeddings that's generated from Wikipedia using open-source models. We will produce three distinct versions of the dataset with embeddings sized at 1024, 4096, and 8192 dimensions to help us get more realistic and robust vector search algorithm benchmark scores across different dimensional requirements.
Current manual evaluation methods for LLM-based AI applications are unsustainable and resource-intensive. While using LLMs as judges offers a promising alternative, evaluation frameworks provide structured tools that harness LLMs for automated assessments, offering metrics, benchmarks, and scalable testing methodologies that overcome manual evaluation limitations. Gemini's integration with evaluation platforms lacks comprehensive documentation compared to providers like OpenAI and Anthropic. Analysis of ten major frameworks reveals scattered integration, decentralized documentation, and limited implementation guidance for Gemini, creating barriers for developers and researchers. I propose a three-pronged approach to improve Gemini support across evaluation frameworks. First, adding native Gemini support to select frameworks (DeepEval and TruLens) that already offer direct integrations for other providers. Second, enhancing both abstraction-layer and framework-specific documentation, covering API key management, model configuration, and multimodal capabilities. Third, creating framework-specific guides in Python and TypeScript that demonstrate effective combination of library features with Gemini APIs. For code deliverables, I will develop native Gemini integration for DeepEval and TruLens frameworks, with pull requests submitted to respective GitHub repositories. Documentation deliverables include enhanced guidance for abstraction layers (LiteLLM, LangChain, LlamaIndex) focused on Gemini configuration, and framework-specific documentation covering API management, configuration options, multimodal integration, and evaluation best practices. Tutorial deliverables will include comprehensive guides for all ten analyzed frameworks, with Python implementations for all frameworks and TypeScript for supported ones, step-by-step configuration instructions, advanced usage examples highlighting Gemini's multimodal capabilities, and error handling techniques.
Summary: AI-Powered Chat Automation for Rocket.Chat Problem: Setting up automated workflows in Rocket.Chat usually requires manual scripting, which can be time-consuming and complex for non-technical users. Solution: We will build an LLM-powered app that lets users create and run automated workflows using simple natural language. It will: 1. Allow users to define triggers (e.g., "when someone mentions 'help'") and actions (e.g., "send them a message asking what they need help with"). 2. Break down complex commands into steps and execute them safely. 3. Prevent harmful actions with built-in safety checks. 4. Let admins manage and review all workflows. Deliverables: 1. A app that understands natural language commands. 2. Multi-step workflow automation 3. Error handling and admin controls for safety. 4. Customizable automation for any Rocket.Chat server. This project will save time by making chat automation easy and accessible for all users.
This project aims to create a web-based tool that can make user interfaces from "Thing Descriptions" (TDs) - a term from the Web of Things standards. Unlike other tools out there, ours will handle multiple things at once so users can monitor and control their IoT setups from one screen. The project will develop the WoT UI component library that can be reused and Eclipse Thingweb platform with an entirely device driven dashboard and UI components that can be dynamically adjusted based on device capabilities without any coding/manual selection needed for new devices. Deliverables include: A Handy Web App to Create User Interfaces A Library to Power UI for WoT Real-time interaction using HTTP and WebSocket protocols. A UI-Affordance Mapping documentation guide mapping TD elements to UI components. AI-powered suggestions for TDs with unknown or irregular structures.
Last year, work on GestureCap focused on developing an Arduino-based latency measurement system and optimizing the pipeline for low-latency performance. Additionally, OSC-based communication between Python and PureData was implemented to enable real-time audio processing. Building on this foundation, this project will introduce AI-driven gesture recognition using lightweight, GPU-accelerated models to improve accuracy and responsiveness. The gesture-music mapping will be expanded to support both trigger-based events (e.g., drum beats) and continuous controls (e.g., pitch, reverb, and panning), allowing for more expressive musical interactions. The project will also integrate GestureCap with MaxMSP and Ableton Live, making it more accessible to musicians and producers. Further improvements will focus on latency reduction, real-time inference optimizations, and testing OpenPose-based feature extraction as a potential enhancement. If time permits, a visualizer for live gesture-music interaction will also be developed. These advancements will refine GestureCap’s capabilities, ensuring a more responsive and expressive experience for users.
Secure Paste Custom Actions on iOS aims to resolve a major limitation in Flutter apps on iOS by enabling developers to add custom actions to the native edit menu. Since iOS 15 introduced the "Secure Paste" privacy notification, Flutter apps have suffered from usability disruptions. This project proposes to build a comprehensive Dart API and a native iOS (Objective-C) implementation that bridges Flutter's widget system with iOS’s UIEditMenuInteraction. It will provide seamless clipboard interaction while respecting privacy constraints. The solution will include: a customizable Dart API, platform channel communication between Flutter and iOS, native menu logic in Objective-C, full test coverage, and detailed developer documentation. If accepted, this project will significantly improve Flutter’s iOS platform maturity and empower developers to enhance user experiences with context-aware edit menus.
The bayesplot package is a key tool in the Stan ecosystem for visualizing Bayesian analysis results. However, it currently lacks specialized visualizations for predictive checks of discrete and categorical outcomes, and some existing plots could be improved for better usability and accuracy. These limitations hinder researchers' ability to thoroughly validate their models against real-world data, particularly for discrete and bounded data types. This project aims to expand and refine bayesplot by implementing new visualization techniques based on recommendations from Säilynoja et al. (2025). Deliverables: New plotting functions integrated into bayesplot (e.g., ppc_density_bounded, calibration plots). Enhanced versions of existing functions (e.g., rootograms, residual plots). Comprehensive documentation with examples and guides. Unit tests for robustness and edge cases. Blog posts tracking progress and showcasing new features. By the end of the project, bayesplot will offer a more versatile and user-friendly toolkit for Bayesian model checking, strengthening its role in the Stan community. These improvements will empower researchers to diagnose and communicate model fit more effectively, advancing Bayesian workflows across disciplines like machine learning, statistics, and data science.
This project aims to improve the reconnaissance scan features of Nettacker while also addressing the issues caused by high parallelization. Additional deliverables include better output types and improved test coverage.
This proposal aims to improve existing data import tools, such as better conversion of Slack to Zulip features when importing Slack's data export, as well as starting new data import and mirroring tools for Discord.
This project addresses the computational cost of full Geant4 calorimeter shower simulation by constructing compact, readout-segmentation–independent point clouds from Geant4 steps for fast-simulation training. I implemented and evaluated three clustering methods—DBSCAN, CLUE-style approach, and an axis-aligned Grid Clustering approach—together with a validation suite (longitudinal/radial profiles, containment radii, raw moments, energy closure) and a side-by-side 3D viewer. Across parameter scans, DBSCAN and CLUE achieved compression but degraded radial structure (and DBSCAN showed mm-level longitudinal drift), whereas the Grid method preserved longitudinal and radial profiles while remaining fast and angle-robust.
The Bene Fundraising Platform on the Ergo blockchain uses a unique token system (APT/PFT) to support fundraising projects. This project will improve the Ergo version by adding support for multiple tokens, including stablecoins, and bringing it in line with the EVM version. Main goals include: adding stablecoin and multi-token support, integrating multiple wallets, building a referral system with protection against misuse, creating a project analytics dashboard, and improving the user interface and accessibility. These updates will make Bene easier to use, more powerful, and more welcoming to new users, helping grow the Ergo fundraising community.
The project aims to enhance and upstream the existing wpanusb and bcfserial drivers in Linux to meet upstream quality standards. Currently, several key driver operations are missing, such as listen before talk and setting frame retries. Preliminary sketches exist but are non-functional pending corresponding changes in the Zephyr firmware side. The project will also add the ability to read the device's permanent extended address, support generic radio configuration (including band, channels, and power levels), and provide comprehensive querying of radio properties. In addition, improvements are required to accommodate upcoming features like management frames and active/passive scanning. This effort is aimed at enabling seamless interoperation between the radio and Zigbee protocols while ensuring support for both 2.4GHz and SubGHz operations, ensuring that the solution meets upstream quality standards and is maintainable over the long term.
This project focuses on integrating a Ternary Content Addressable Memory (TCAM) module into the Chipyard SoC framework to enable high-speed parallel matching operations crucial for domains like networking and pattern recognition—functionality not natively supported in Chipyard. The integration employs a dual approach: first, incorporating the Verilog-based TCAM IP (generated via OPENTCAM) as a memory-mapped I/O (MMIO) peripheral by wrapping it in a Chisel BlackBox, connecting it to the subsystem using TileLink, and exposing TCAM operations like write and search through memory-mapped registers accessible from software; and second, integrating the TCAM as a RoCC accelerator by reusing the BlackBox within a custom LazyRoCC module, mapping Rocket’s RoCC interface to TCAM operations via funct fields of custom instructions, and enabling instantiation through a dedicated WithTCAMRoCC configuration in the Rocket tile.
JavaScript lacks a general purpose and cache efficient multi-dimensional array API for operating on large datasets of high-dimensionality. To address this limitation, stdlib provides ndarray API for creating and manipulating multi-dimensional arrays. This project aims to extend those efforts by achieving full API parity between stdlib ndarrays and built-in JavaScript arrays. This work will include implementing functions such as concat, find, flat, includes, indexOf, reduce, some, none, any, with, sort and others to facilitate high-performance data transformation and manipulation. For performance intensive operations such as searching, sorting and reductions, this project will develop C implementations and corresponding Node.js native add-ons for accelerated performance. These enhancements will allow JavaScript developers to work with multi-dimensional arrays with the same ease as working with built-in arrays and thus expand JavaScript's capabilities for scientific and numerical computing.
This project aims to significantly enhance the search experience within documentation generated by Documenter.jl. The current client-side search encounters difficulties accurately indexing and retrieving results for Julia-specific syntax, leading to suboptimal relevance. Furthermore, issues like unnecessarily large search indices and repeated listing of the same location hinder usability. This proposal outlines a plan to address these shortcomings by: 1) Implementing search performance benchmarks (relevance, precision/recall, index size) to measure and guide improvements. 2) Developing and integrating an improved, Julia-aware tokenizer within the search JavaScript. 3) Optimizing the index by investigating and removing redundant data, ensuring unique location entries are stored, and adding warnings for potentially problematic large indices. 4) Creating comprehensive developer documentation for the enhanced search system. The outcome will be a more robust, accurate, and efficient search capability for users navigating Julia package documentation.
This project develops an open-source 5G cell scanner utilizing GNU Radio and SDR hardware to passively detect and decode 5G signals, extracting key broadcast data, including the MIB (Master Information Block) and SIB1 (System Information Block 1). It provides real-time monitoring, multi-band scanning, and improved usability for researchers and network analysts. 1. Python-based DSP pipeline for 5G synchronization and decoding. 2. GNU Radio out-of-tree (OOT) module for real-time Software-Defined Radio (SDR) processing. 3. Visualization and logging tools for 5G network analysis. 4. Comprehensive documentation for ease of use and future development. This tool will expand GNU Radio’s 5G capabilities, supporting open-source wireless research and advanced cellular network analysis.
OpenMRS currently utilizes Hibernate Envers to track database changes, offering robust audit capabilities. However, the system lacks a user-friendly way for administrators to view, filter, and interact with audit logs. This project proposes building a backend module that exposes audit log data in a meaningful way through a legacy UI-based audit log viewer, REST endpoints for external access, and an extension that tracks read operations on selected entities—something Envers does not support out-of-the-box. The solution will include: 1. Audit log retrieval and visualization within the legacy UI 2. Custom tracking of read operations using Hibernate event listeners or an alternative approach 3. Filtering capabilities (by user, date, and entity) 4. RESTful APIs for external access to audit data 5. A well-documented architecture, extensible to other OpenMRS modules By the end of this project, OpenMRS administrators will have a comprehensive and usable interface to audit system activity, improving transparency, traceability, and compliance with healthcare data standards.
Currently, the autocomplete feature in the p5.js Web Editor is facilitated by CodeMirror and a custom-built static hinter (as given in the structured JSON file, p5-hinter.js). The hinter suggests all the available functions, and while it is helpful, the current implementation lacks context-awareness, and that can be confusing, especially for beginners. In this project, I will improve the autocomplete by making it context-aware which means it will only show suggestions that make sense based on where the cursor is in the code. I will also add autocomplete for user-defined functions, variables, and classes, and support features like jump-to-definition and renaming. This will make the editor more beginner-friendly, intuitive, and efficient to use.
This project aims to enhance gprMax's user support capabilities by integrating advanced Language Model (LLM) technologies. The core goal is developing an AI-driven chatbot capable of interpreting natural language user instructions and generating accurate, ready-to-run gprMax input files. The project is scheduled for completion within a 12-week timeframe, divided into clear milestones ensuring systematic progression and iterative refinements. The final deliverables will great improve the user experience of gprMax and laying foundation for future AI integration.
This project focuses on making Wagtail.org more energy-efficient by adapting its content based on the environmental impact of the user's location. When electricity is being generated from more carbon-intensive sources, the website will automatically adjust its behavior, such as serving lighter images or reducing animations to lower its energy usage. The goal is to create a more sustainable browsing experience while maintaining performance and accessibility. Along the way, insights will be gathered to help guide other developers in building low-carbon websites using Wagtail.
This project implements a Maven plugin that generates accurate OpenAPI 3.0 specifications for any OpenMRS module by introspecting REST resources at build-time. The plugin leverages the OpenMRS test context within a forked JVM to discover all resource handlers, automatically detecting every available property including those accessible through custom representations, without requiring any code annotations or modifications. Key deliverables include the functional Maven plugin with comprehensive schema introspection, generated OpenAPI specifications with full representation support and developer documentation, enabling automated API documentation for OpenMRS implementations.
The MBDyn Python Preprocessor simplifies the creation of input files for MBDyn simulations by bridging Python programming with MBDyn's input format. This project enhances the simulation workflow through two key improvements. The first transforms the YAML-based Rotorcraft Parametric Modeling tool into Python classes within the preprocessor, creating an object-oriented framework for rotorcraft modeling. The second integrates the preprocessor with Blendyn (a Blender add-on), enabling users to create, edit, simulate and visualize MBDyn models entirely within Blender. These enhancements make MBDyn modeling more intuitive and accessible while providing a seamless workflow for both pre- and post-processing tasks.
This project prototypes a new set of Cargo plumbing commands, aligned with Rust’s 2025H1 project goals. Plumbing commands are lower-level, less opinionated commands to interact with the Cargo API. By focusing on lower-level functionality, this project aims to better support Cargo users’ specific processes and use cases. Moreover, the introduction of these new Cargo plumbing commands aims to fix current issues of existing commands and improve their integration.