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.
<p>DeepFalcon is an ultra-fast non-parametric detector simulation package. This project aims to extend DeepFalcon by adding functionality for Graph Normalizing Flows (GNFs) to it.</p> <p>While previous work on DeepFalcon has focused on other Deep Generative Models and Graph Neural Networks to simulate detector event reconstructions, we aim to use GNFs to create generative models for graph structures that can better approximate detector reconstructions.</p>
<p>The goal of the project is to integrate FHIR Standard API’s that will help in exchanging healthcare information electronically. This will allow the import and export of patients data while maintaining FHIR standards. Currently we have Patient and Encounter resources with some services. My goal with this project is to complete existing implementation and integrate as many resources as possible and start implementing search capability.</p>
<p>The libvirt library provides a stable API for managing platform virtualization utilizing different hypervisors or virtual machine monitors. Each hypervisor requires its own driver inside libvirt, but not all API functions are supported by all drivers or hypervisors yet. In addition to the standard drivers, there is a supplementary fake driver, called test driver, designed to let applications test against libvirt with fake data and not have any effect on the host. As of today, there are still a lot of API functions not implemented by the test driver. The goal of this project is to expand the API coverage of the test libvirt driver.</p>
<p>The Data Repository Service (DRS) API provides a generic interface to data repositories so data consumers, including workflow systems, can access data in a single, standard way regardless of where it’s stored and how it’s managed. The main goal of this project is the generation of a prototype server for DRS API that follows the current draft specification of the DRS</p>
<p>Processing Language Server focuses on creating a Language Server Protocol (LSP) implementation for Processing Programming Language. PDE is currently built using Java and using custom components of the Swing Framework, which is ~ deprecated. The long term goal of Processing is to replace this with a JS based IDE to bring in more contributors and to make building UIs simple. While planning on building such IDE, LSP is of significant importance for any language that the IDE relays on. Since Processing is the targeted Programming language, it’s quite important to build a Language Server Protocol for the same. This shall act as a benchmark for all the crucial activities of the IDE such as auto-completion, go-to-definition, hover-insights and so on. LSP will also help in easy and seamless integration of the above functionalities in any editor such as Atom, VScode, etc.</p>
<p>As the student who took the initiative together with the mentors Jeremiah Foster and Gunnar Andersson during GSoC 2018 for Voice Command on IVI Systems, we were successful in developing a conceptual approach called “VCIVING” and implementing a voice command and response system on PCs. This was called “EmulationCore” and was capable of playing music and finding a location on the map based on user’s speech. Somewhat primitive methods were adopted to refine exact information from the user’s speech those of which have a greater possibility to fail. I intend to work on enhancing the speech recognition and the implementation of VCIVING on GDP.</p>
<p>Currently in Pitivi, the UI for configuring an effect applied to a clip is generated dynamically. Also, as of now we simply use a Numeric widget to specify RGB values of color correction effects.</p> <ul> <li>This project will provide a mechanism to add custom UI for effects.</li> <li>A custom UI for 3 point color balance effect using color wheel widget to adjust shadows, midtones and highlights would be implemented to improve user work-flow.</li> </ul>
<p>As a member of the sage-dynamics community, researchers have compiled a wishlist for algorithms and functionality they would like added. I would like to shorten the wish list for us.For my project I will be completing some desired additions to SAGE from the Sage Dynamics Wiki. I will implement Well’s Algorithm, strengthen the numerical precision in cannonical_height, as well as implement reduced_form for higher dimensions.</p>
<p>The proposed project deals with the development of two Drupal modules, for storage and visualizations of Complex Biological Graph. Both the proposed modules will be able to integrate along with Tripal and can be used by the Scientific community for the construction of genetic website, publish genomic data. The primary aim of the project is to optimize querying by using a Graph Database and also generate a good User Interface.</p>
<p>From Visual(Deep) Neural Network Builder, users can drag and drop layers and other computational units and they can create neural networks as they wish. They can test,train the networks and this builder will be integrating with WSO2 Machine Learner.This will be the front end for the WSO2 Machine Learner.</p>
This project focuses on extending the existing React based Mifos X Web App to make it a strong candidate for the primary Mifos X web interface. The application will be enhanced using React and TypeScript, with improved integration to the Apache Fineract backend through the OpenAPI TypeScript client. This project will result in fully functional multi-step workflows and the implementation of missing or partially developed features. The final outcome will be a stable web application with feature parity with the existing Mifos X platform.
Borg stores each deduplicated chunk as a separate file. On SFTP, that's one network round-trip per chunk. Pack files bundle N chunks into one file. A TOC sits in the trailer, and the pack gets a SHA256 derived pack_id. The client caches a small index per pack locally. Borg compact handles cleanup by mark-and-sweep to find dead chunks, then rewrite sparse packs into dense ones. Deliverables: pack file write and read support in borg2, borg compact with pack rewriting, incremental per-pack index cache, and encrypted pack support.
This project focuses on making SymPy's assumptions system more robust, efficient, and reliable through a three component architectural upgrade. The first component introduces a backend solver for Equality Logic and Uninterpreted Functions (EUF) directly into the SAT engine. The second component upgrades the Linear Real Arithmetic (LRA) solver to safely handle complex domains, irrational boundary limits, and symbolic infinities (like oo) without failing while also refactoring the core ask dispatcher to resolve many recursion loops during query evaluation. Last component, the project will replace thousands of hardcoded static unit tests with dynamic, property-based tests using the Hypothesis library. By automatically generating complex mathematical trees to find hidden bugs, these enhancements will smooth out the rough edges of the assumptions module, making it a significantly faster and mathematically stronger.
This project develops linear-scale attention vision transformers for CMS End-to-End jet classification and proxy mass regression on 8-channel 99.1% sparse detector images. Five architectures were trained and benchmarked under identical conditions during the test task: XCiT with domain MAE pretraining (AUC 0.9237, 2.42x S/sqrt(B)), XCiT with ImageNet pretraining, XCiT from scratch, a custom L2ViT combining local window and linear global attention built from scratch, and Swin-T as the softmax baseline. All models use a dual head for joint classification and proxy mass regression. 11 analysis studies were completed beyond standard benchmarking: channel importance (Ch0 and Ch3 critical, AUC collapses to 0.47 without either), channel pair interactions, Grad-CAM saliency, t-SNE feature representations, layer-by-layer CKA similarity, data efficiency (5x labelled data savings), jet multiplicity analysis, energy threshold study (detector implications for zero-suppression), S/sqrt(B) significance curves, adversarial robustness, and calibration with temperature scaling. Key findings: domain MAE pretraining outperforms scratch by +0.125 AUC and leads to fundamentally different representations (CKA 0.008-0.015 at every layer), linear attention beats softmax by +0.063 AUC under identical conditions, and block 1 of the pretrained model alone exceeds the fully trained scratch model in feature separation. The summer work extends this to multi-class classification (quark/gluon/top/W/Z) with per-class channel importance ablations, token pruning exploiting 99.1% sparsity, sparse convolution baselines, hybrid CNN+ViT architectures, and comparison against ParticleNet and JEDI-net on the same held-out test set.
The Weather Routing Tool is a system designed to find optimal shipping routes between two geographic points, taking into account weather data, local boundaries, and other environmental factors. The tool currently assumes constant ship speed and has a basic genetic algorithm for route optimization. This project aims to improve the routing logic by: Introducing ship speed optimization along a route Enhancing the genetic algorithm for more effective route modeling Improving the usability and extensibility of the tool for end-users Project Deliverables 1. Ship Speed Optimization Goal: Allow optimization of ship speed along the route instead of assuming it constant. Approach: Modify the Isofuel algorithm to use speed as a minimization criterion. Define a new scoring function: score = (distance / time) - alpha * remaining_distance Enable this feature via configuration: ISOCHRONE_MINIMISATION_CRITERION = "speed" 2. Genetic Algorithm Enhancements Goal: Improve the existing genetic algorithm’s efficiency and flexibility. Planned Improvements: Add multi-objective cost function support (e.g., fuel, speed, wear & tear). Enhance crossover logic with: Multi-point crossover Uniform crossover Adaptive crossover Improve mutation strategies and population initialization. Leverage the pymoo library with custom configurations. 3. Tool Usability Improvements Goal: Make the Weather Routing Tool more modular, user-friendly, and extensible. Key Features: Dynamic Python file injection using CLI: python cli.py -f config.template.json custom.py Load modules dynamically with: load_file_as_module(path: str) -> types.ModuleType Command-line utility packaging: Tool installable via PyPI (pip install wrt) Usage: wrt -h Rationale: Enable users to add custom boat models and configurations without modifying core code.
This project addresses the challenge of data drift in high-energy physics, where evolving detector conditions cause AI models for data quality monitoring to become obsolete. Fully retraining these models is computationally expensive and can lead to catastrophic forgetting of past information. To overcome this, we are implementing a parameter-efficient continual learning framework using a deep autoencoder. Our method involves freezing the large, pre-trained encoder-decoder core and only fine-tuning the lightweight outer convolutional layers to adapt to new data runs. This approach allows the model to adjust to new data distributions without altering its core knowledge, maintaining high classification performance (measured by ROC AUC) while reducing the number of trainable parameters by over 60%, thus offering a robust and sustainable solution for real-time data quality assurance.
Accurately classifying particle collisions—whether distinguishing quark- from gluon-initiated jets or isolating Higgs events from complex backgrounds—requires models that capture both fundamental conservation laws and the inherent geo- metric invariances of the detector environment. This work presents a hybrid self-supervised framework that leverages physics-informed deep learning to ad- dress these challenges. Our approach utilizes a masked transformer encoder to learn conservation of momentum and energy by reconstructing masked particle features, while a Lorentz-Equivariant Geometric Algebra Transformer (L-GATr) encodes the space-time symmetries present in collision events. By integrating these complementary embeddings into task-specific cross-attention decoders, the framework effectively fuses physical and geometric information to enhance event classification and mass reconstruction, thereby advancing the performance and interpretability of machine learning in high-energy physics.
Fast and accurate simulation of high-energy physics experiments is crucial for advancing our understanding of fundamental particles and forces in nature. However, the traditional method of obtaining such simulations using Monte Carlo methods is computationally expensive. Recently, machine learning techniques such as generative modeling have been proposed as an alternative approach to providing simulation results. In particular, diffusion models have emerged as highly accurate, but the standard formulation of the diffusion process suffers from slow inference speeds. This project aims to accelerate the inference of diffusion models by exploring recently proposed techniques in the field, such as distillation, continuous-time diffusion formulations, or more efficient architectures. The methods developed in this project will be integrated into the Geant4FasSim repository, enhancing its capabilities for fast and accurate simulation of high-energy physics experiments.
The Active Segmentation Platform for ImageJ (ASP/IJ) aims to provide a general-purpose workbench that allows biologists and other domain experts to access state-of-the-art techniques in machine learning to achieve excellent image segmentation and classification. This project enhances the functionality and user experience of ASP/IJ by updating its user interfaces and developing a new visualization and reporting panel, exposing the Weka advanced visualization and analysis functions.
This project enhanced Gradle Build Server for Android development by implementing composite build support, improving Java Home handling, and providing support for Android Java projects. These enhancements significantly improve the development experience for Android developers using Text Editors/IDEs that utilize the Build Server Protocol (BSP). By bridging the gap between Android Studio and Gradle Build Server, this project contributes to a more productive development workflow. This is a start to let developers benefit from the powerful capabilities of Android Studio within their preferred development environments.
cargo-semver-checks is an amazing tool in the Rust ecosystem that automatically detects violations in Semantic Versioning (semver) using lints. However, it cannot currently be controlled with enough granularity, which means that not every library can currently adopt this. With this project, users of cargo-semver-checks will be able to configure the program at the per-lint level, making it much more feasible for libraries to use this tool and increase adherence and trust in SemVer.
The use of superpages in modern hardware and operating system yielded important performance benefits. Unfortunately, allocating superpages can often fail due to a high rate of external fragmentation in physical memory. Thus external physical memory fragmentation in modern operating systems has once again become a serious issue. This project aims to design and implement several anti-fragmentation mechanisms for the FreeBSD kernel. At the time of writing, the FreeBSD kernel has no memory compaction mechanisms or any way of quantifying physical memory fragmentation. The first major goal is implementing several fragmentation metrics for the page allocation code. The second major goal is designing and implementing a runtime physical memory compaction mechanism. When invoked, this mechanism would identify sparsely populated and fragmented physical memory regions and rearrange the pages to eliminate any "holes" in the region. Aside from explicit invocation, a background daemon would monitor the relevant fragmentation metric and trigger the memory compaction mechanism should the value drop below a certain threshold. The last goal will introduce a passive anti-fragmentation mechanism for the demand paging mechanism. The key idea is to enhance the page placement code to use contiguous spans of pages instead of allocating random pages.
The users of IHR fall under the category of students, educators, professionals, network operators, policymakers, and researchers. The goal of this proposal is to visualize the internet resources through Internet Yellow Pages (IYP), which acts as an end-to-end solution for stakeholders to work on. Currently, IYP has data related to AS, IXP, Peer ASes, Sibling ASes, Countries, URLs, etc. The data can be visualized through tables, charts, and graphs (wherever possible) so that users can access much information. For example, a lot of information can be visualized about an AS including its Prefixes, IXP, Peer ASes, Sibling Ases, etc. I propose to develop a search page for the Internet Yellow Pages which is self-exploratory. The search page would let the user select either AS, Country, and IXP and correspondingly render tables, charts, and graphs to visualize the entity.