Explore Projects

<p>The project will increase Image Sequencer capability while simultaneously demonstrating the ability to process satellite images. The general approach is to develop Image Sequencer as a user interface for openCV.js and openCV.js as computer vision processing engine for Image Sequencer. Technical objectives include: 1) streamlining satellite processing capability via image sequencer functions, 2) enabling/extending opencv applications for Image Sequencer and 3) demonstrate daily satellite environmental analysis over a 3 month period.</p>

SQLancer is a tool to automatically test Database Management Systems (DBMSs) in order to find bugs in their implementation. That is, it finds bugs in the code of the DBMS implementation, rather than in queries written by the user. SQLancer has found hundreds of bugs in mature and widely-known DBMSs. DQE is a method for systematically detecting logical errors in UPDATE and DELETE queries. Integrating Published Testing Approaches into SQLancer makes up for the limitation of existing tools (such as PQS, NoREC, TLP) that only target SELECT queries, and significantly improves SQLancer's vulnerability detection capabilities.

<p>Apache Nemo (incubating) is a data processing system that supports various deployment characteristics, by easily customizing translation of a dataflow program into a physical execution plan. Supporting RESTful API and web interface can help Apache Nemo to:</p> <ul> <li>Provide easy way to inspect compiler passes and runtime modules for developers.</li> <li>Provide intuitive way to introspect the behaviour of Nemo stack and to make reasonable decision on their configuration for application writers and cluster operators.</li> </ul>

Description: Vitess is a database clustering system for horizontal scaling of MySQL. One of the key goals of Vitess is to emulate MySQL behavior even while running multiple MySQL instances so that ORMs and frameworks work seamlessly. Vitess has its own in-built SQL-parser which it uses to understand the query and represent as structs for further processing. As of now, a lot of MySQL functions are not parsed correctly and result in syntax errors. Parsing for a lot of the newer features in MySQL 8.0 is also missing. The task of the mentee would be to add parsing support for such functions and features.

The goal of the project is to implement distributed tracing in order to trace individual calls between multiple backend components. The implementation can be broken down into the following stages: Infrastructure: Deploy an OpenTelemetry colector, and a Tempo tracing backend; Instrumentation: Instrument backend components in order to send events to the collector; Tracing: Stitch events together into traces by propagating trace context; Polishing: Iterate the generated events in order to improve the observability of the system; Documentation: Document how to deploy the required components, how to add/modify generated events, and how to visualize the data.

I read the source code of libssh and I notice that there’re many switch-case in the sftp server code, while the implementation to handle different IO events in the server part is based on a call-based model, which has a quite clear logic. And sftp is a subsystem in libssh which can be established through a channel, as a subsystem having many sub-message type, it’s better to use a callback based method to implement a sftp server. Therefore, I’m going to modify the sftp part in libssh and write a sftp server with the api in libssh to achieve this goal.

GPUs and CUDA have revolutionized computing beyond graphics rendering, becoming essential in various fields due to their parallel processing capabilities. Clad, a Clang plugin, provides automatic differentiation for C++ functions by modifying Abstract-Syntax-Tree using LLVM compiler features. While Clad supports reverse-mode differentiation of CUDA kernels, data-race conditions are very frequent and considerably slow down the execution. Thread Safety Analysis helps detect data races to optimize performance by potentially reducing atomic operations in generated code.

<p>The project has been under constant development for almost a year now. In the past one year, the app has been reworked to compile with the current Android ecosystem. Users can view release information by scanning a barcode, search information about artists, releases, release groups,labels, recordings, instruments, and events , view collections, tag their audio files (to a limited extent) and donate to the MetaBrainz Foundation via PayPal. The application has been released as a beta version on the Play Store for testing purposes. This project aims to make the application stable and robust.</p>

Ganga is a tool for composing, running, and tracking computing jobs across a variety of backends and application types. Ganga primarily runs as a command-line tool and has an IPython-like prompt. My proposal lays out a plan to add a large language model to the Ganga prompt. It discusses options like cloud serving, local serving, and the challenges that come with each. It also addresses whether a rag system would make sense for the project at this juncture. Finally, I suggest a timeline breakup where I first focus on implementing a local solution followed by a cloud one.

The aim of this project is to develop a comprehensive learning resource for developers working with the Gemini Python SDK. This would consist of a range of documentation, interactive code samples and an easy to use pre-written package that demonstrate cover a range of topics which can be used to efficiently manage long content, one of the most common challenges when building with large language models. It will provide an overview of techniques such as batching, chunking and handling various media types (such as text, audio and video) in addition to highlighting best practices for techniques such as caching and error handling.

Chorus 2 is the default Web Interface for Kodi. It has been implemented on Coffeescript, Backbone, Marionette, etc. Kodi now has a project in development called Elm-Chorus which is the re-implementation of Chorus 2 in Elm functional language as it provides easy maintainability of code, error reporting, etc. In Elm-Chorus, the basic layout and functionality from Chorus 2 has already been implemented but it still lacks some features, styling, and layout pages. The aim of this project is to 'finish the new web interface' in Elm by adding the remaining functionality and UI from Chorus 2.

Meshery (https://meshery.io) is a self-service engineering platform, Meshery enables collaborative design and operation of cloud native infrastructure. [Meshery Catalog](https://meshery.io/catalog) content represents a schema-based description of cloud native infrastructure. Catalog content need to be portable between Meshery deployments as well as easily version-able in external repositories. - Expected outcome: Rebuild the Meshery Design System so that it provides the open source building blocks to design and implement consistent, accessible, and delightful product experiences.

<p>The proposal for GSoC - Zulip aims to improve the overall experience of the Zulip web app by means of implementing full stack features such as <strong>send later feature</strong>, <strong>remind about message feature</strong>, <strong>clear status text after time feature</strong>, <strong>deny guest users access to other users</strong>, etc. I would like to see Zulip able to cater the needs of its growing user base by implementing features essential for a smooth and high-performance team chat system.</p>

<p>One of the important aspects of searches for new physics at the Large Hadron Collider (LHC) involves the identification and reconstruction of single particles, jets and event topologies of interest in collision events. The End-to-End Deep Learning (E2E) project in the CMS experiment focuses on the development of these reconstruction and identification tasks with innovative deep learning approaches.</p> <p>This project focuses on the integration of E2E code with the CMSSW inference engine for use in reconstruction algorithms in offline and high-level trigger systems of the CMS experiment.</p>

This project implements key components of KEP-2170, introducing the Kubeflow Training V2 API. Specifically, it focuses on creating the TrainingRuntime and ClusterTrainingRuntime for the JAX and TensorFlow frameworks, built upon the Kubernetes JobSet API. These runtimes will serve as blueprints for model training (including LLMs) within cloud-native ML pipelines. This abstraction allows Data Scientists and MLOps Engineers to easily reuse standardized runtimes and launch training jobs, particularly via the SDK, without needing deep knowledge of underlying Kubernetes complexities.

This project improved Gemini AI integration in Roo Code VS Code extension, serving 800,000+ developers. Key deliverables included real-time Google Search grounding for live web research, progressive model migration eliminating configuration errors, and explicit context caching design reducing token costs by 75%. Additionally, fixed critical streaming race conditions in citation handling and designed cost-effective solutions making AI coding assistants accessible to individual developers and small teams.

The aim of this project is to gain deeper insights into driver behavior, identifying potential risk factors, and developing effective interventions to improve transportation safety. Key tasks include literature review, data preprocessing,cognitive load prediction for distraction, visual & manual distraction detection,driving style classification, crash risk prediction and gaze analysis. The expected outcomes of this project is accurate driver behavior analysis and risk assessment based on data and evaluation of driver performance suggestion of interventions for distraction based on gaze analysis

Open Climate Fix recent Cloudcasting model is able to predict clouds movements very accurately but for end user it’s hard to see or visualize the results, this project aims to create a new dashboard similar to quartz-frontend that is already up and running, to visualize results from the cloudcasting model the dashboard will be build using Next.js and MapBox as base map layer, Sentry to log errors, and a minimal fastapi server written in python to preprocess zaar data and serve it to the frontend, this solution will be a proof of concept for the new MVP

Problem: Introduce New CRD ShardingSphereChaos to ShardingSphere. Plan: 1. Design chaos based on ShardingSphere in a production environment 2.Implement it to out environment by operator.define the status and spec of the corresponding crd based on the actual chaos implemented (considering its duration, running state). The chaos can be managed by writing corresponding logic code in reconcile. 3.make it automatic and do a lot chaos experiments to improve system availability. Result: 1.chaos CRD and controller. 2.chaosEngineer theory about ShardingSphere 3.automatic chaos engineer

Currently the documentation for the Syntax of the various SQL’s that Polypheny supports has text based format which proves rather hard to read. The goal of the suggested project by Polypheny was to create a mini system which can convert the BNF like syntax in the markdown into a more visual representation in the form of railroad diagrams. With this I would like to extend the project into a large Project to apply the implementation of this Framework to the existing Jekyll Process of rendering so that it shows up as required on the website. I would also like to pursue possible extensions to improve the syntax completion and highlighting in the query console in the Polypheny-UI.

The Kruise Game Controller Manager currently stores cache information for network plugins in memory, resulting in its single-replica deployment. From a business stability architecture perspective, it is necessary to migrate the cache out of memory to enable Kruise Game to transition to a multi-replica deployment. Besides, the current webhook certificate is self-signed by the controller. When the number of copies is more than one, an unauthenticated error will occur, so this needs to be modified. Expected Outcome: 1. Support multi-replicas deployment 2. Support webhook certificate signing using cert manager

Time series forecasting is paramount in many domains, including finance, healthcare, energy, and climate science. This project suggests incorporating deep learning-based forecasting models—Informer, TCN, and DeepAR—into the aeon/tookit. The objectives are to construct an efficient and scalable framework for forecasting that accommodates top-performing models, is compatible with Aeon’s data management, and provides stable training, evaluation, and documentation. Through simplifying the availability of advanced forecasting software, the project would make it easier to utilize the toolkit to assist researchers with streamlined time series analysis.

<p>As everyone knows, TensorFlow is the best open source platform for Deep Learning. Instead of focusing on basics like alternatives which lag behind TensorFlow on visualization capabilities, documentation and not only datasets but also models, TensorFlow engineers develop the next-generation technologies. It is acceptable to say that TensorFlow shapes today’s technology. Therefore, I am eager to work on projects offered by TensorFlow in order to enrich mine in open-source software development by learning from the best and to apply my knowledge into TensorFlow which is the last word in machine learning libraries.</p>