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><strong>TensorFlow Lite Flutter Support Suite</strong> is a cluster of three projects TfLite Flutter Plugin, TfLite Flutter Helper library, Flutter example ML app along with a tutorial.</p> <p><strong>TfLite Flutter Plugin</strong> provides TfLite JAVA like dart API to perform inference in flutter apps. The plugin is built on the top of the existing dart bindings and binds to TfLite C API using dart:ffi.</p> <p>View <a href="https://github.com/am15h/tflite_flutter_plugin/" target="_blank">am15h/tflite_flutter_plugin</a> & <a href="https://pub.dev/packages/tflite_flutter" target="_blank">tflite_flutter on Pub</a></p> <p><strong>TfLite Flutter Helper library</strong> aims at providing an interface to easily use high-level objects involved in Image, Text, Audio processing with TfLite Flutter plugin, and related utility functions in addition to similar functionality as the TensorFlow Lite Android Support Library.</p> <p>View <a href="https://github.com/am15h/tflite_flutter_helper/" target="_blank">am15h/tflite_flutter_helper</a> & <a href="https://pub.dev/packages/tflite_flutter_helper" target="_blank">tflite_flutter_helper on Pub</a></p> <p>A comprehensive and detailed <strong>Flutter example ML app</strong> demonstrating the combined efficient usage of the TfLite Flutter plugin and TfLite Flutter Helper package together will be published with a detailed tutorial.</p> <p>View <a href="https://github.com/am15h/object_detection_flutter/" target="_blank">am15h/object_detection_flutter</a> & <a href="https://medium.com/@am15hg/real-time-object-detection-using-new-tensorflow-lite-flutter-support-ea41263e801d" target="_blank">blog</a></p>
EcoGrid Intelligence: AI-Driven Climate Resilience Analysis for Global Energy Infrastructure on Liquid Galaxy is an AI-driven geospatial analysis system designed to proactively identify global energy infrastructure at risk of operational failure due to severe climate events such as heatwaves, droughts, and wind anomalies. It addresses the problem of critical infrastructure and meteorological data being siloed, which prevents stakeholders from achieving a unified, spatial understanding of risk. The solution dynamically fuses live climate anomalies with global energy infrastructure maps to create an interactive analytical canvas on the Liquid Galaxy platform. The EcoGrid Intelligence system follows a layered architecture: Data Sources and Risk Assessment: Climate Data: Streams live and forecasted climate anomalies, including temperature deviations, precipitation extremes, and wind intensity, from the Open-Meteo API. Infrastructure Data: Geospatial coordinates, generation type (hydro, nuclear, solar), and operational capacity are sourced from the Global Power Plant Database. Risk Model: A deterministic model computes a Climate Vulnerability Score (CVS) by evaluating localized anomaly intensity against the plant type's operational sensitivities. AI Orchestration: Google’s Gemini models act as the system’s brain to autonomously determine region-relevant climate anomalies, trigger API calls, interpret the resulting CVS metrics, and structure the output for spatial rendering. Visualization: The results are converted into Keyhole Markup Language (KML), generating elements like Placemark, LookAt, and FlyTo for native, synchronized rendering and camera guidance on the Liquid Galaxy multi-screen rig. Outputs: Primary displays render immersive climate anomaly heatmaps and color-coded nodes indicating vulnerability levels, while secondary screens display dynamically generated HTML content and AI-driven insights.
<p>Kuadrant's MCP Gateway is an Envoy-based gateway for Model Context Protocol (MCP) servers. It currently handles authentication and coarse-grained authorization via AuthPolicy/Authorino, but has no built-in mechanism for tool-level authorization, e.g., allowing agent A to call `add` and `subtract` on a math server while agent B can only call `subtract`. The kube-agentic-networking SIG is defining a standard AccessPolicy CRD for exactly this. MCP Gateway is well-positioned to implement it because its ext_proc already parses MCP request bodies and exposes tool metadata via headers (`x-mcp-method`, `x-mcp-toolname`, `x-mcp-servername`). This project will design how AccessPolicy maps to Kuadrant's AuthPolicy/Authorino, then build an experimental implementation covering CRD definition, a controller that translates tool-level authorization rules into Authorino configuration, identity model integration (starting with OIDC), `tools/list` response filtering so callers only see tools they're authorized to use, and CEL-based authorization expressions. The mentee will work closely with mentors on the design doc, pair on implementation decisions, and engage with both the Kuadrant and kube-agentic-networking communities.</p><p>- Expected Outcome:</p><p> - Design document covering AccessPolicy-to-AuthPolicy/Authorino mapping, body parsing strategy (ext_proc vs Envoy MCP filter), identity model choices, `tools/list` filtering, and CEL-based authorization</p><p> - AccessPolicy CRD with status conditions and CEL validation rules</p><p> - Controller that watches AccessPolicy resources and generates/updates AuthPolicy resources with authorization rules based on `x-mcp-toolname` headers</p><p> - OIDC identity source support, with consideration for ServiceAccount and SPIFFE identity types</p><p> - `tools/list` response filtering so callers only see tools they are authorized to invoke</p><p> - CEL-based authorization rule support (e.g., `request.mcp.tool_name.startsWith("read_")`)</p><p> - E2E tests covering allow/deny scenarios for tool-level access control</p><p> - Documentation guide in `docs/guides/`</p><p><br></p>
<p>Meshery is the open source cloud native manager that empowers platform engineers to design and operate infrastructure. As infrastructure complexity grows, the need for intelligent assistance becomes critical. This project focuses on developing and enhancing a dedicated AI Adapter and AI Connections for Meshery. This adapter serves as the bridge between Meshery’s core orchestration engine and various Large Language Models (LLMs). The goal is to enable "Natural Language to Infrastructure" capabilities, allowing users to describe their architectural intent (e.g., "Deploy a highly available Kubernetes cluster on AWS with Prometheus monitoring") and have Meshery auto-generate the visual topology and configuration manifests. The intern will work on decoupling the AI logic from the core platform, allowing users to "Bring Your Own Key" (BYOK)—supporting both cloud-based providers (OpenAI, Anthropic) and local inference runners (Ollama, LocalAI).</p><p><br></p><p>Recommended Skills:</p><p> - Proficiency in Golang (Go) is essential, as Meshery’s backend is written in Go.</p><p> - Familiarity with MCP Servers, REST APIs, LLM APIs (OpenAI, Vertex AI), local inference servers (Ollama).</p><p> - Basic understanding of Kubernetes, Docker, and Infrastructure-as-Code (IaC) concepts.</p><p> - Experience with REST, GraphQL, and gRPC.</p><p> - Nice to have: Experience with React (for frontend integration in Meshery UI).</p><p><br></p><p>Responsibilities:</p><p> - Co-design and implement the interface for the AI Adapter in Go to communicate with the Meshery Server.</p><p> - Implement support for connecting to local LLMs (via Ollama) to ensure data privacy for users who cannot send infrastructure data to the public cloud.</p><p> - Improve the "System Prompt" and context-window management to feed the LLM relevant data regarding Meshery Models (schema definitions) so the AI generates valid infrastructure configurations.</p><p> - Write unit and integration tests to ensure the reliability of the adapter.</p><p> - Create user guides on how to configure the adapter with different AI providers.</p><p>- Expected Outcome:</p><p> - A fully functional AI Adapter (or Connection) integrated into the Meshery ecosystem.</p><p> - Demonstrable capability for users to swap between at least two different LLM providers (e.g., OpenAI vs. a local Llama 3 model).</p><p> - Implementation of a feature where natural language queries result in a rendered design.</p><p> - Merged pull requests (PRs) including code, tests, and documentation.</p><p><br></p>
<p>Kyverno recently completed two CNCF assessments: a security assessment by the CNCF TAG Security & Compliance group and a General Technical Review by the CNCF TOC Project Reviews subproject. Together they produced a set of findings spanning documentation, threat modeling, network policies, global context cache bounds, API server authentication, SAST tooling, API stability and non-goals, UX/adopter research, webhook cert issuance/rotation guidance, "safe-mode" / temporary disable patterns, SLOs/SLIs and alerting/runbooks, dependency lifecycle and SCA workflows, third-party notices, and the security response process. The findings are tracked in umbrella issues for each assessment. In this mentorship, the mentee will work through the open findings from **both** the TAG Security & Compliance assessment and the General Technical Review, propose and implement fixes across the Kyverno codebase, docs, and Helm charts, and help close out both assessments. Work includes implementing cache bounds for the Global Context, restricting Global Context access in namespaced policies, adding API server request authentication for the admission webhook, generating sample/network-policy templates and a CLI command to produce a Kyverno NetworkPolicy, integrating SAST tooling (e.g. semgrep, Nancy) into CI, updating the threat model and architecture diagrams, documenting core CRDs/API stability, webhook cert rotation, safe-mode/incident playbooks, SLOs/SLIs and reference dashboards, the SCA/dependency lifecycle workflow, and improving the security documentation on kyverno.io.</p><p><br></p><p>Expected Outcome:</p><p> - Resolve the open findings tracked in [kyverno/kyverno#15335](https://github.com/kyverno/kyverno/issues/15335) (TAG Security & Compliance assessment) and [kyverno/kyverno#15473](https://github.com/kyverno/kyverno/issues/15473) (General Technical Review assessment).</p><p> - Implement Global Context cache bounds and access restrictions for namespaced policies (kyverno/kyverno#15359).</p><p> - Add admission webhook authentication of requests from the API server.</p><p> - Refresh the threat model (including CLI and other deployment options) and the architecture diagram to separate logical and physical components.</p><p> - Document core CRDs/APIs and per-API-group stability policy, webhook cert issuance/rotation guidance, recommended "temporary disable" / safe-mode patterns, SLOs/SLIs with reference dashboards/runbooks, and the end-to-end dependency/SCA workflow.</p><p> - Update the Kyverno security documentation: fix the audits page links, document risks of external data lookups, link to published security advisories, and link the current security response process.</p><p> - Add tests covering the new behaviors and document the changes in the Kyverno docs site.</p><p><br></p>
<p>libavutil is being expanded with a modern way of handling frequency transformation. The project aims to significantly increase the speed of this code on a selected architecture by writing the transformations in assembly.</p>
I will implement FengWu-GHR model into OpenClimateFix graph_weather repository. This model is the first to be able to predict weather at the 0.09° scale, despite training mostly on ERA5 dataset.
The main aim of this project is to link generic Python functions to SU2 to facilitate data-driven fluid modeling while simultaneously enabling thermochemical data interpolation to enhance the accuracy and efficacy of these fluid models.
<p>In order to bridge the gap between industry and new technology, this project will supply governance models which are relevant to Privacy-Preserving Machine Learning. This project in particular will be focusing on recommending an incident response model.</p>
The project aims to enhance the Statistics package in the GNU Octave by adding Generalised Additive Models (GAMs) Regression and kNN classification functionality. These features will allow users to perform better statistical analyses and modeling with ease of interpretability.
The project should deliver an AI-enabled backup camera application for AGL using Fluorite. It captures multiple camera feeds via Gstreamer, stitches them into a top-down bird's-eye view, and a YOLOv8 model for object detection. The model should run on TFLite.
This proposal regards mapping histogram data to its underlying function; both of which can be represented as sequences and hence a transformer can and will be used to represent the mapping. This project involves both generating functions and histograms to use as a dataset and implementing and training a transformer to perform the mapping.
Train an image segmentation model. Do both segmentation and obstacle distance calculation in it. Burn the model into Luxonis AI camera. Then use the on-camera ESP32 to send obstacles’ relative location to ArduPilot through MAVLink. In this way, we can use that camera for object avoidance on boats.
<p>Task would be to get the data from Elasticsearch as a stream and model it as a graph in order to provide insights on the fly. Once we map the data stream into a graph model, we can use the high performing graph algorithms to cluster the data and send through different nodes for further processing.</p>
<p>Text in a video often conveys information that is not easily expressed otherwise. This project will implement and compare the various existing state-of-the-art scene-text-transcription models and run inference of the best model on Java using legacy Java bindings of Tensorflow.</p>
<p>LabelLab allows the user to upload images and label them using bounding boxes and polygons. Currently, there is no way for the user to build models using the labelled dataset which they have created. The goal of the project is to create a pipeline which allows the users to train classification models using the images which have been uploaded and labelled on LabelLab.</p>
Creating a source generator for Corvus.JsonSchema, which adds the code produced by the tool to the user's source so that it can be compiled. By leveraging JSON Schema as a unified source of truth for data modelling, teams can generate code for their respective data models while adhering to a standardised data structure defined by JSON Schema.
<p>This is a project that develops a simulator consisting of an executor model abstracted based on statistics and a scheduler model that are executed like actual running with manually managed timeline. Directly, this project can increase the reliability of nemo. Indirectly it can contribute to speed up future projects and be a springboard to future projects.</p>
<p>The aim of this project is to create a framework that can help people create a web based demo out of their machine learning code and share it. Others can test the model without going into the implementation details. Usually testing models by other people involves a lot of preparation and setup. This project aims to cut that down.</p>
In this proposal, I proposed my idea of deploying a Kubernetes cluster consisting of docker containers of ML data set along with ML model code. Training in this model can be automated using CI/CD pipeline Jenkins. Dataset and code could be fetched from Github by developer repo by triggering the repo of developer at a certain amount of time
Sktime provides a user-friendly interface for a range of forecasting algorithms and time-series functionalities. To enhance its capabilities further, we can integrate deep learning models from pytorch and huggingface. This integration will significantly extend Sktime’s functionality by incorporating advanced deep learning architectures, foundational models, and global forecasting capabilities.
<p>We come up with models that give the effective and efficient visualizations as recommendations to the given input. Current visualization tools require the user to manually select attributes and analyze the data. For someone who has limited time and domain,this gets challenging if there are millions of attributes to derive insights. To overcome this problem, we automate the process with deep models</p>
<p>The goal of my project is to create a data model close enough to the reality, manage to interface it with an existing gateway simulation (that would be the Kura simulator of the Kapua project) and finally make the data available through the Kapua Cloud platform. Afterwards, a GUI interface is to be implemented to give the possibility to visualise data and change the parameters of the model as desired.</p>
<p>Fabrik is an online collaborative platform to build, visualize and train deep learning models via a simple drag-and-drop interface. It allows researchers to collaboratively develop and debug models using a web GUI that supports importing, editing and exporting networks written in widely popular frameworks like Caffe, Keras, and TensorFlow.</p>