Explore Projects

This Project aims to address the challenge of articulating and transcribing musical ideas effectively through the utilization of generative AI technology. The proposed changes consists of developing a user-friendly tool that allows individuals to record or upload their musical concepts and receive potential transcriptions in Music Blocks code format. Leveraging Javascript expertise, musical theory/composition knowledge, and large language models (LLMs), the plan is to create an intuitive interface for recording/uploading audio samples, implement a server-side LLM service for analysis, and establish an API for seamless data communication. Upon acceptance of this proposal the deliverables will include an in-app interface, server-side LLM service, and API for enhanced musical creation and transcription assistance.

Milvus currently lacks a Debezium source connector, meaning data can flow into Milvus (via sink connectors) but not out of it into the broader data ecosystem. This critical gap forces users into manual workarounds for data replication, auditing, and prevents the creation of hybrid pipelines that join vector search results with relational data. The plan is to implement a Hybrid strategy focused exclusively on the future-proof Milvus 2.6+ StreamingNode architecture. This approach will leverage two separate sources to guarantee a complete and ordered change stream: Woodpecker Streaming Service (WAL): Used as the only source for all Data Manipulation Language (DML) events (inserts and deletes). Etcd Watch: Used as the cleanest source for all Data Definition Language (DDL) events (schema changes), which simplifies parsing and aligns with Debezium's schema history mechanism.

<p>The Unified Push Server Database is currently based on MySQL/PostgreSQL, and the data is all stored in one schema. Database updates are handled by a JMS broker. However, when this system is put under pressure with a high level of users and an intense quantity of push metrics it is unable to perform optimally. The reality here could be a poor user experience with slow message delivery, dropped messages or the possibility of service unavailability. As messages begin to exceed millions, this will become a bottleneck for the system and one which needs to be addressed and engineered with scalability in mind.</p> <p>This project will investigate the possibilities of using different types of databases and methods to process large volumes of data. This will be a Proof of Concept that will aim to research Apache Kafka and Apache HBase as possible solutions to improve the real-time experience of the push metric of the AeroGear Unified Push Server.</p>

Implement TransformStream backpressure evaluations and compression stream polyfills in the C++ Workerd runtime.

<p>Storing, sharing and accessing the vast volumes of data being generated in modern research and healthcare settings is extremely daunting. With the Data Repository Service (DRS) schema, the GA4GH has developed an API specification that provides access to data stored in the cloud, via a uniform experience. The DRS schema is part of the GA4GH Cloud Work Stream, which allows users to execute FAIR workflows on FAIR data in the cloud. However, it currently does not allow data to be dynamically added, which is a limitation to some use cases for federated computing.</p> <p>The goal of this project is to develop an open-source, generic (i.e., not tied to any specific data provider), distributable and highly reusable DRS microservice implementation with various, diverse and unique use cases in the operationalization of the GA4GH Cloud Work Stream, such as enabling user upload in client applications or on-site caching, temporary/intermediate storage functionality for execution endpoints and as a lightweight and maintainable demo service for GA4GH Cloud interoperability testing.</p>

This project aims to enhance cBioportal's analytical capabilities by integrating Variant Allele Frequency (VAF) visualization into the Plots Tab and OncoPrint views. Currently, in Plots Tab we can visualize mutation types of particular gene across different cancer types but it is not possible to visualize VAF data for different samples across different cancer types. It is very crucial for researchers to understand mutation clonality and prevalence across cohorts through VAF. Similarly in OncoPrint view we cannot add tracks for VAF heatmaps across different genes. The proposed solution has two components. For the Plots Tab, we will introduce VAF (linked to a specific gene) as a selectable numerical data type of mutation for plot axes. Existing plot components will then automatically render Box-Scatter or Scatter plots. For OncoPrint, we will implement gene-specific VAF heatmap tracks displayed below the main alteration grid. VAF heatmaps corresponding to primary displayed mutations with a dedicated 0-1 color scale.

<p>The Charmil Framework will help in resolving the problems in apache Kafka examples and improve Apache Kafka CLI experience by providing a number of popular Kafka CLI’s as one single HOST CLI with others installable as plugins.</p> <h4>Deliverables</h4> <ul> <li>Provide example applications that can be used to showcase command-line experience.</li> <li>Improve base framework (charmil) to Install a plugin CLI in the Host CLI</li> <li>Build and deploy registry with different versions of Kafka specific CLIs</li> <li>GitHub repo/pages based remote index servers for plugins which will contain docs, metadata, etc.</li> </ul> <p>The user will import charmil in their CLI, which will provide them all the extensions out of the box. The user will be able to install specific plugins with the install command. After installing the extension, users will be able to use the features of that plugin without writing code for that particular feature which that plugin provides.</p> <p>A GitHub repo will be used as the index server for all plugins. The repository will hold a directory of JSON / YAML files. Each describes a single plugin. This directory is called Plugin Manifest.</p>

<p>The highlyUnstable branch of robocomp has an advanced rcmanager, which aids for creation of the xml files in a simple and illustrative way, through a GUI. However, there are a few problems associated with it, like cluttering of large components graphs, unavailability of GUI for editing components, no support for graph panning, no support for component debugging, as well as insufficient documentation of the new tools. I intend to solve these problems through my GSoC project, whose elaborately described solutions can be found in my proposal.</p>

cBioPortal hosts 400+ cancer genomics studies whose clinical metadata is wildly heterogeneous — treatment fields alone have 24+ variants. The mentor's team built MetaHarmonizer, a 4-stage ML pipeline using sentence-transformer semantic matching and ontology resolution (NCIT, MONDO, UBERON), but it has no web interface, no audit trail, and no cBioPortal-ready export. This project wraps the pipeline in a production-ready curator dashboard: upload a raw metadata CSV, review and correct automated column mappings with confidence scores and top-5 alternatives, validate ontology term assignments, monitor quality KPIs, and export a cBioPortal-compatible TSV with a full JSON audit report. I have already built a working full-stack prototype (React + FastAPI) integrating the real SchemaMapEngine — processing 141 columns in 2.6 seconds — with 20+ REST endpoints, an on-demand LLM rematch endpoint, and an automated evaluation pipeline on the provided eval set. Deliverables: (1) production-grade backend with PostgreSQL persistence and role-based auth, (2) polished React dashboard with all needed curator pages, (3) formal ML benchmarking report, (4) CI/CD pipeline and (5) full documentation and cBioPortal deployment guide.

Description:Kruize Autotune is a Performance Tuning Tool for Kubernetes. . It uses Hyper Parameter Optimization to tune the language runtime layers of a given application. System component metrics can give a better look into what is happening inside them. Metrics are particularly useful for building dashboards and alerts. Kubernetes components emit metrics in Prometheus format. This format is structured plain text, designed so that people and machines can both read it.Following are four common methods of hyperparameter optimization for machine learning in order of increasing efficiency: 1.Manual 2.Grid search 3.Random search 4.Bayesian model-based optimization The aim of Bayesian reasoning is to become “less wrong” with more data which these approaches do by continually updating the surrogate probability model after each evaluation of the objective function. At a high-level, Bayesian optimization methods are efficient because they choose the next hyperparameters in an informed manner. The basic idea is: spend a little more time selecting the next hyperparameters in order to make fewer calls to the objective function. In practice, the time spent selecting the next hyperparameters is inconsequential compared to the time spent in the objective function. By evaluating hyperparameters that appear more promising from past results, Bayesian methods can find better model settings than random search in fewer iterations.So,this project aims at tuning kubernetes applications using bayesian optimization.

The project aims to enhance the cBioPortal platform by generating a list of recommended default genes for each OncoTree code, which are often used for molecular classification of specific cancer subtypes. This will improve the effectiveness of the cBioPortal database by prioritizing the display of relevant genes to each disease subtype. To achieve this goal, a Large Language Model (LLM) such as GPT-4o, will be fine-tuned with prompt engineering and Retrieval-Augmented Generation (RAG) techniques to provide domain-specific knowledge and context. The goal is to train the LLM to connect relevant genes to each OncoTree code. Approach: LLM Selection: GPT-4o is a strong candidate due to its versatility and performance in various NLP tasks. Prompt Engineering: Develop efficient prompts for the selected LLM to generate gene lists. RAG Component: Implement a RAG component to provide context to the LLM and enhance accuracy using e-utilities to retrieve curated literature from NCBI GeneReviews and ClinVar. Validation: Cross-reference the generated gene lists with expert-curated resources like ClinGen and variant frequencies in cBioPortal, and COSMIC. Evaluation: Continuously monitor the model's performance using validation sets and adjust hyperparameters as needed. Deployment: Integrate the solution into the cBioPortal platform such as sorting variants on patient view pages by relevant genes first. Conclusion: Completion of this project will significantly improve the usability of cBioPortal, providing users like researchers and clinicians with relevant gene and pathway information based on the cancer type they are exploring.

OpenDAL is a data access layer that allows users to easily and efficiently retrieve data from various storage services in a unified way. oftp can expose OpenDAL power in FTP way that allow users to access storage services via FTP protocol.

<p>Solution which will accelerate a process of creating conference event</p>

<p>SoundCom is an mobile application which can transmit data through sound waves.</p>

<p><strong>VisualCircuit</strong> allows users to program robotic intelligence using a visual language which consists of blocks and wires, similar to electronic circuits. Currently it's a desktop application released for Linux based on IceStudio which uses AngularJS. The aim of this project is to make a similar web application so that users do not have to do any installation and use newer frameworks on the front-end since EOL for AngularJS is in December 2021. The requirements of the project are:</p> <ol> <li>Make a web app that has all the functionalities as the original desktop app.</li> <li>Add a django based wrapper to the existing backend so that it can communicate with the web-app.</li> </ol>

<p>RoboComp is an open-source Robotics framework providing the tools to create and modify software components that communicate through public interfaces. It is based on DSL technology. In this proposal we seek to improve the robustness of the generated code using model checking techniques and good software engineering practices. The student will have to extend the current code generation tool, robocompdsl, to include verification points, asserts, parameter range control, etc independent of the specific functionality of the component. We expect that with this improvement the new generated components will be less prone to accidental crashes and easier to debug and maintain.</p>

<p>I think about how to improve the RoboComp system by adding visual detection mechanisms to the mobile robots. This provides to the robot all the information needed to predict its short-term environment and react better than a robot without this technology (it could look less human, it couldn't know about its probable near future so it couldn't react to it). That is to say, if a person passes in front of the robot, probably, it can't know it to react and border the person, because it is an unpredictable event. But, with this technology, if a human crosses ahead, the robot detects a change in its environment and it can change its path to not collide with the human. I have used RoboComp before and I know that it is a good software that can be greatly improve with this system.</p>

<p>A desktop GUI application that can obtain task lists from various sources (Debian Bug Tracker's iCalendar feed and Github iCalendar, CalDAV servers like DAViCal and the Redmine REST API) and display a Kanban board, as used in Scrum/Agile.</p>

<p>As more efforts have been made to Spark SQL, there are more demands on providing more information and metrics through Spark UI in an effective way. Specifically, users and developers want to (1) browse information for every individual Spark SQL sessions, (2) get continuous updates from jobs created by structured streaming, and (3) have an easy and intuitive way to debug runtime issues. In this work, we aim to make efforts to response the above three demands by adding features, improvements, and related infrastructures to Spark UI for Spark SQL and structured streaming.</p>

This project mainly aims to improve the overall code quality of O3 by enhancing the component test coverage. It also expects a well written documentation explaining the test strategy and best practices which other developers can follow for the future development of O3. Following this documentation will ensure that the quality of o3 is maintained and developers can have the confidence in their developed features.

Vitess uses arewefastyet to automatically benchmark its codebase and ensure no performance regression is introduced. The mentee will have the responsibility of continuing the UI that was previously created using React/Vite. - Expected Outcome: The expected outcome is to continue working on the Frontend UI that was developed during the 2023-summer term, that includes adding an admin UI, adding a feature to ensure the consistency of the results, improving the overall UX of the website, and add new pages to improve the scope of arewefastyet.

Previous iterations of GSoC combined with recent community contributions have helped TensorFlow catch up with the SOTA in image classification. However, object detection in TensorFlow is yet to bridge that gap. This project aims to contribute modular components used in modern object detectors to KerasCV. Moreover, some deliverables are proposed outside the main project to further enhance the usability of KerasCV for Computer Vision Engineers. Finally, the use of these modules will be depicted through the creation of YoloX-Darknet53 in a demo notebook and a keras.io tutorial.

<p>Open event orga server offers the great solution for Organizing event. There are great tools and features like sessions, speakers, ticket management with the ability to export and provide API of the event. These all offer a great solution to organize an event with ease.</p> <p>There are certain features which can help Organizer to connect more with audience and speakers. The organizer should be able to receive feedback on​ the sessions and query regarding the event. This project focuses on same with providing such features for organizer to connect more with attendees, past audiences, speakers.</p>