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.
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 Model" (BYOM)—supporting both cloud-based providers (OpenAI, Anthropic) and local inference runners (Ollama, LocalAI). Expected Outcome: 1. A fully functional AI Adapter (or Connection) integrated into the Meshery ecosystem. 2. Demonstrable capability for users to swap between at least two different LLM providers (e.g., OpenAI vs. a local Llama 3 model). 3. Implementation of a feature where natural language queries result in a rendered design. 4. Merged pull requests (PRs) including code, tests, and documentation.
<p>The proposal is inspired by the idea G on RedHen's GSoC 2016 idea page. The main purpose is to develop models and code helping domain experts to research on art history. It includes five main tasks:</p> <ol> <li>Training a classifier for recognizing a limited number of known visual symbols in images of religious artworks. VGG-net based fully convolutional network models will be trained for this task. Multi-scale detection, non-maximum suppression and bounding box regression will be implemented. </li> <li>Training a detector for localizing person heads and hands in images of art with various styles. CNN models will be trained based on a state-of-art head detection method.</li> <li>Building a module identifying a few certain hand gestures in images of artworks based on task 2. With the cropped detection result of task 2, a classifier based on VGG-net will be built.</li> <li>Developing a tool generating a frontal view of a face given a artistic portrait with a side view of the face based on existing face frontalization code. </li> <li>Developing initial research code for analysis of color style changes over time of Christian paintings by machine learning.</li> </ol>
This project addresses the challenge of limited and imbalanced datasets in coronary artery calcium (CAC) analysis, particularly for rare high-risk cases such as high Agatston scores. Existing augmentation methods rely on heuristic or purely data-driven approaches that lack physiological realism. We propose a physics-informed generative framework for synthetic CAC data creation. Building on a validated atlas-based registration pipeline, the project models plaque formation as a stochastic process governed by latent hemodynamic factors. A reduced-order physics-informed neural network (PINN) will estimate endothelial shear stress (ESS), which will guide a stochastic differential equation (SDE)-based model to simulate realistic plaque growth patterns. The system will generate anatomically and physiologically plausible calcium distributions, with controllable parameters such as Agatston score, vessel territory, and lesion phenotype. Deliverables include a modular Python-based generation pipeline, a synthetic dataset of 300–500 scans covering long-tail cases, and a comprehensive evaluation using statistical, structural, and expert-based validation metrics. This work shifts CAC data augmentation from heuristic placement toward physics-driven simulation, enabling more robust and generalisable machine learning models for medical imaging.
This project addresses the issue of translating complex marine biodiversity data into actionable knowledge for ecosystem management in Florida. It makes use of existing biodiversity datasets (OBIS, GBIF) and ecosystem variables (e.g., temperature) to develop a collection of reproducible Jupyter Notebooks powered by Python packages like pyOBIS, GeoPandas, and machine learning packages. These notebooks are dedicated to showcasing the top environmental inquiries in areas like the Florida Keys, Rookery Bay, Tampa Bay, and the Gulf of Mexico covering trends in Rice's Whale, birds, and seagrasses. The solution combines species occurrence records and ecosystem information to generate visualizations, spatial models, and biodiversity metrics. Interactive maps, temporal trend plots, and model-based habitat predictions are products. Constructed with reproducibility and open science principles in mind, the tools will assist resource managers, researchers, and decision-makers in conservation planning. Deliverables include: 5 modular Jupyter Notebooks (e.g., habitat models, trend predictions), Reusable data pipeline functions (data ingestion, cleaning, analysis), Public Docker-supported GitHub repository, Easy-to-read documentation and automated testing framework. The project facilitates environmental decision-making through data science and encourages collaboration through open, reproducible tools.
The proposal aims to enhance Eclipse Winery, a web-based environment for modeling OASIS TOSCA topologies, by implementing basic security measures. Currently, Winery focuses on providing a graphical editing environment for modeling application topologies. However, the proposal suggests adding a Language Server Protocol (LSP) provider for OASIS TOSCA YAML files. This would allow Winery to support the latest version of the OASIS TOSCA standard and handle dynamic type additions. The expected outcomes of the project include: 1-Developing an LSP server that supports the latest OASIS TOSCA 2.0 standard. 2-Ensuring that the LSP server is aware of newly introduced types and makes them available in the service template. 3-Enabling the LSP server to handle multiple files. 4-Demonstrating the functionality of the LSP server using compatible IDEs such as Eclipse Theia, Visual Studio Code, or Eclipse IDE. 5-Writing clean and maintainable code, with a preference for using modern Java (version 21 or later). The proposal indicates that the project's size is estimated at 350 hours and rates its difficulty as medium. Overall, the proposal aims to improve the security and functionality of Eclipse Winery by implementing an LSP server for OASIS TOSCA YAML files, thereby providing a more robust and comprehensive modeling environment.
<p>Current Red Hen Lab’s Audio Pipeline can be extended to support speech recognition. This project proposes the development of a deep neural-net speech to text module for the pipeline, based on the paper Deep Speech. The aim is to use both audio and visual modalities for achieving speech recognition.</p> <p>The initial goal is to extend current Deep Speech model (audio only) to Red Hen lab's TV news videos datasets. The next goal is to develop a multi-modal Speech to Text system (AVSR) by extracting visual modalities and concatenating them to the previous inputs.</p> <h5>Project Outline</h5> <p>I plan to develop four versions of Speech to Text during 12 weeks time.</p> <ul> <li>Version 1: Rewriting the Deep Speech model to support audio inputs from Red Hen Lab datasets. </li> <li>Version 2: Improving results using either N-gram model or a spell check system</li> <li>Version 3: Extracting visual features, concatenate them with Audio features and modify Deep Speech’s Input</li> <li>Version 4: Improving results using the same approach as in 2, tracking actual speaker’s lips, etc.</li> </ul>
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 current KubeEdge platform represents device states using three statuses: Desired, ObservedDesired, and Reported. The device states displayed on the platform are entirely reliant on the Mapper, which collects and reports data from the device side. However, due to limitations in the Mapper implementation, physical device malfunctions, network delays, and potential network attacks, the device states shown on the platform may not accurately reflect the actual state of the devices. In the KubeEdge platform, if applications depend on device states for decision-making, such inconsistencies in state representation may lead to undesirable outcomes. Therefore, this project aims to design a device state anomaly detection framework for KubeEdge. By exploring the causal relationships among device states, the framework will establish lightweight anomaly detection capabilities and provide a comprehensive toolchain encompassing data collection, model training, real-time anomaly detection, and results visualization. Expected Outcome: 1. A general-purpose device anomaly detection framework that supports user-defined detection algorithms 2. A complete technical design document including model selection, training procedures, and detailed architecture diagrams for both training and online detection components 3. A machine learning model and corresponding anomaly detection algorithm capable of capturing causal relationships among device states, trained and tested using standard frameworks...
Spatial transcriptomics methods identify tissue domains by clustering spatial gene expression patterns, but current approaches select genes using variance-based heuristics such as highly variable genes (HVGs). Many HVGs reflect donor-specific noise rather than genes that truly drive spatial organization, which leads to poor generalization when models are applied across tissue slices from different donors. This project addresses this limitation by introducing CauST, a framework for identifying causally important genes for spatial domain identification. CauST evaluates each gene through in-silico gene knockout experiments, measuring how much removing the gene changes the learned spatial embedding produced by a spatial graph neural network. To ensure robustness, the method then applies a cross-slice invariance criterion, selecting genes whose effects are both strong and stable across multiple donors. The resulting causally invariant gene set is used to retrain spatial domain identification models, enabling improved cross-donor performance without modifying the underlying model architecture. The project will deliver a production-quality open-source CauST Python package, integration with multiple spatial transcriptomics backbones (STAGATE, GraphST, and SpaGCN), a multi-dataset benchmark evaluating cross-slice robustness, visualization tools for interpreting causal gene importance, and tutorial notebooks and documentation to support adoption by the spatial transcriptomics community.
This project focuses on developing a new machine learning approach called Quantum Model-Agnostic Meta-Learning (Q-MAML) specifically for high-energy physics (HEP) experiments like those at the Large Hadron Collider (LHC). These experiments produce massive amounts of data that change over time, making it challenging to train machine learning models effectively. Q-MAML uses quantum computing techniques with variational quantum circuits to enable "few-shot learning." This means the models can quickly adapt to new data with very little retraining, which is crucial for the dynamic environment of particle colliders. The goal is to improve tasks like identifying particle collision patterns (event topology classification) and finding rare signals. The project will develop a modular, open-source Q-MAML workflow that is compatible with both quantum simulators and actual quantum hardware, utilizing platforms like Qiskit and PennyLane. The performance of Q-MAML will be rigorously benchmarked against classical MAML and standard variational quantum algorithm (VQA) classifiers on physics-inspired tasks. These benchmarks will include detailed evaluations of training curves, adaptation behaviors, and resource analysis. The deliverables will consist of a fully documented codebase, interactive Jupyter notebooks for ease of use, and reproducible benchmarking tools. This work aims to establish a practical foundation for hybrid quantum-classical learning in scientific applications, particularly within the field of high-energy physics.
Detection of Intonational Units- An overhaul of AuToBI aims to refurbish the AuToBI system first proposed by Andrew Rosenberg in 2009, with today’s state-of-the-art algorithms and models so as to cater to the compute, speed and accuracy requirements that are characteristic to today and the future. The main idea is still the automatic detection and classification of prosodic events (Rosenberg’s thesis considers two events of interest pitch accents and phrase boundaries, but we can target to cater to a wider array of events), but we will lay our focus on the broad goal of establishing a new baseline by employing some of the more sophisticated MLmethods as of today. One of the fairly less-explored avenues of today is the interpretability of models, and with our work, we hope to set precedent for the future of prosody-detection with a highly interpretable model Goals and Deliverables: Goal 1: Try to incorporate speaking rate and speech rhythm as additional prosodic events. Correlated Goal 1A: Analyse the correlation between the prosodic events of speech rate/rhythm with phrase boundaries. Goal 2: Try to incorporate CTC alignment to do away with disparities due to frame-level alignment. Analyse the benefits of the change. Goal 3: Employ techniques like Tandem DNN-HMMs, RNN-Transducers and WSFTs and analyse the benefits, if any Ambitious Goal 4: Employ state-of-the-art techniques to improve speaker diarization. Can try to generalise speakers to larger groups based on ethnicity/region to employ region-specific analysis.
New openSUSE users face a steep learning curve navigating distribution-specific tools like zypper, YaST, Btrfs/Snapper, and systemd. This project builds a locally-running AI onboarding assistant that pairs a Small Language Model (SLM) with a Retrieval-Augmented Generation (RAG) pipeline over official openSUSE documentation. The assistant detects the user's actual system context (desktop environment, pending updates, failed services, filesystem layout) and grounds every answer in real documentation and real system signals. It runs fully offline after setup, respecting user privacy. A working proof of concept already exists with local SLM inference (TinyLlama 1.1B via llama-cpp-python), a ChromaDB-backed RAG pipeline, system context detection, CLI and Web UI interfaces, and a benchmarking framework. Deliverables: (1) production-ready SLM with tiered model support, (2) enhanced RAG pipeline with 4+ documentation sources and tuned retrieval, (3) expanded system context detection covering Btrfs/Snapper, networking, GPU, and locale, (4) jeos-firstboot module for first-boot integration, (5) systemd service with socket activation, (6) RPM packaging via OBS with subpackages for core, docs, and model, (7) security hardening and prompt injection defenses, (8) comprehensive testing across hardware tiers, and (9) blog posts and user documentation.
CAR T-cell therapy is a form of cancer immunotherapy that engineers a patient’s T cells to recognize and eliminate malignant cells. Although highly effective in leukemias and other hematological cancers, this therapy faces significant challenges in solid tumors due to the complex and heterogeneous tumor microenvironment. CARTopiaX is an advanced agent-based model developed to address this challenge, using the mathematical framework proposed in the Nature paper “In silico study of heterogeneous tumour-derived organoid response to CAR T-cell therapy,” successfully replicating its core results. Built on BioDynaMo, a high-performance, open-source platform for large-scale and modular biological modeling, CARTopiaX enables detailed exploration of complex biological interactions, hypothesis testing, and data-driven discovery within solid tumor microenvironments. The project achieved major milestones, including simulations that run more than twice as fast as previous model, allowing rapid scenario exploration and robust hypothesis validation; high-quality, well-structured, and maintainable C++ code developed following modern software engineering principles; and a scalable, modular, and extensible architecture that fosters collaboration, customization, and the continuous evolution of an open-source ecosystem. Altogether, this work represents a meaningful advancement in computational biology, providing researchers with a powerful tool to investigate CAR T-cell dynamics in solid tumor and accelerating scientific discovery while reducing the time and cost associated with experimental wet-lab research.
The Problem General-purpose Automatic Speech Recognition (ASR) often fails language researchers and learners because it is designed to "clean up" speech. In second-language learner data, it is critical to capture exact production, including disfluencies, attempts at pronunciation, and interjections. Current models often struggle to segment these recordings correctly or erroneously "fix" a student's grammatical and phonological errors. The Solution My proposal implements a robust, three-phase automated pipeline: - Pre-processing: I will implement noise reduction using tools like noisereduce or DeepFilterNet and utilize industry-standard VAD (Voice Activity Detection) such as Silero VAD or Pyannote.audio to segment audio into single-sentence blocks. - Transcription: Using high-fidelity models like faster-whisper, the system will capture all vocalizations with word-level timestamps. This phase focuses on preserving the learner's actual speech patterns rather than idealized grammar. - Post-processing: The pipeline will identify pauses, match transcriptions to target sentences, and use Large Language Models (LLMs) with Constrained Prompting. This allows the system to distinguish between intelligible learner errors (to be preserved) and actual [gibberish] (to be flagged). Key Deliverables - Python-based CLI Tool: An efficient command-line interface capable of transcribing raw MP3 files. - High Accuracy: A system designed to achieve >90% agreement with human transcribers. - Structured Output: Final results formatted as an Excel sheet for easy analysis by researchers and learners.
Executive Summary: Create a Comprehensive FHIR facade Layer For OpenELIS Global. The Challenge: The Synchronization Gap The current architecture utilizes a HAPI FHIR JPA "Sidecar" which functions as a secondary, disconnected data store. This results in: Data Latency: Clinical data is trapped in a "sync queue," delaying real-time care coordination. State Drift: Risks of "Split-Brain" where the LIS database and FHIR store disagree on patient records. Resource Waste: Redundant storage of the same data in two different PostgreSQL schemas, increasing infrastructure costs by ~40%. The Solution: Real-Time Native Facade I propose a Native FHIR Facade embedded directly within the OpenELIS Spring context. This transforms OpenELIS into a FHIR-native server without the need for a secondary database. Core Innovations: HAPI Plain Server Integration: Swapping the heavy JPA engine for a lightweight RESTful controller. Live Resource Mapping: A FhirTransformService that maps OpenELIS Domain Objects (Java 21 Records) to FHIR Resources on-the-fly. Atomic Transactions: FHIR writes (POST/PUT) share the same @Transactional context as the LIS, ensuring 100% data integrity. Technical Implementation Stack Language: Java 21 (leveraging Pattern Matching for cleaner transformation logic). Engine: HAPI FHIR (Plain Server Library). Mapping: Manual POJO transformation or MapStruct for high-performance throughput. Validation: Native HAPI FhirValidator to ensure compliance with regional Implementation Guides (IGs).
The current state of the asynchronous I/O on the JVM involves using separate selector threads to manage I/O events through system functions like epoll or kqueue. Although frameworks like Netty and NIO2 make this approach seem reasonable, the use of separate event loop threads is not without cost. In high-RPS services, these threads can compete for CPU time with worker threads executing application tasks, leading to resource contention and potentially significant performance loss. In contrast, Cats Effect offers a different threading model, integrating I/O and application tasks into a single runtime system, reducing contention an enhancing overall efficiency. io_uring, the key component of this project, offers two major advantages: (1) it enables all supported syscalls to become asynchronous from the application’s perspective, allowing for multiple syscalls to be initiated in parallel without blocking the calling thread, and (2) it uses a pair of ring buffers for syscall submission and completion events that are shared between the application and the kernel, dramatically reducing the overhead of calling from userspace into kernelspace. When integrated with Cats Effect, io_uring can significantly improve the efficiency of syscalls, requiring fewer calls into kernelspace during the worker loop. This proposal brings two key innovations together: a new threading model and a new I/O API. By integrating Cats Effects and FS2 with io_uring, we aim to develop an alternative solution to minimize the inefficiencies found in the traditional model, offering a more efficient and performant solution for I/O-intensive applications.
This project addresses the challenge of fast and accurate particle detector simulation in high-energy physics. Traditional Monte Carlo methods are highly accurate but computationally expensive, while existing fast simulation techniques often compromise on physical fidelity. The goal is to develop a model that achieves both high accuracy and significantly lower computational cost. Building on my GSoC 2025 work, I propose to extend a graph-based generative pipeline where particle jets are represented as graphs and processed using Graph Neural Networks (GNNs). The current architecture includes a Distance-based ChebNet encoder, a deterministic latent bottleneck, a physics-aware cluster decoder, and a conditional flow matching model with FiLM conditioning. A key innovation is that generation occurs entirely in a fixed 512-dimensional latent space, making computation independent of the number of particles and substantially more efficient than existing methods. For GSoC 2026, the focus is to refine this pipeline to publication quality. This involves improving key physical observables such as multiplicity, jet mass, and particle momentum using better loss functions and physics-based constraints. I will also integrate a more stable ChebNet and perform ablation studies to validate architectural choices. The project will include rigorous benchmarking against state-of-the-art models like EPiC-FM, EPiC-GAN, and PC-JeDi using metrics such as Wasserstein distance, FPND, and generation time. Deliverables include an optimized pipeline, detailed evaluation and ablation studies, an open-source implementation, and submission to venues like NeurIPS ML4PS.
Electricity grids today serve a far more complex mix of consumers than they were designed for. Data centers, EV charging networks, and industrial facilities participate in demand-response programs, shifting flexible workloads to windows where electricity is cheaper or cleaner. Acting on those signals requires software that collects real-time grid data across multiple regions, forecasts how conditions will evolve over the next several hours, and exposes both through a clean API that developers can query directly. Existing open-source infrastructure for grid signal forecasting covers carbon intensity reasonably well, but critical gaps remain: the data acquisition layer and the API serving layer are not connected, fresh data does not automatically trigger model inference, and signals beyond carbon intensity such as demand, price, generation mix, and grid saturation are not exposed through any unified interface. There is also no interactive frontend that makes regional grid conditions spatially visible. EnergyAPI addresses each of these gaps. The platform is structured across five layers: an automated scheduled data acquisition service pulling from NOAA NOMADS, EIA, and ENTSO-E, an extended relational data model covering all five grid signals, a forecasting pipeline connecting fresh data to existing carbon intensity models while adding SARIMA-based demand forecasting and gradient boosting for price, a Django REST API with a workload scheduling interface that accepts SLOs and returns execution plans, and an interactive web frontend with time-series charts and a choropleth map of EU and US regions.
PROTACs represent a very promising emerging drug modalities. Traditional modalities, small-molecule drugs, for example, or monoclonal antibodies, are limited to certain modes of action, like targeting specific receptors or blocking particular pathways. Due to its ability to target proteins that have been traditionally difficult to target with conventional methods, PROTACs have attracted substantial interest from both academic researchers and pharmaceutical companies, with several PROTACs currently in clinical trials for various indications, including cancer and neurodegenerative disorders. However, the exploration of PROTACs is still relatively in its infancy. With many open research questions, my proposal implements a tutorial, a common feature in DeepChem, to explore how to computational investigate PROTACs. More specifically, I provide a guide towards exploring available PROTAC datasets, developing featurization capabilities, and a multimodal sequence and structure model to infer protein degradation abilities of PROTACs, a clinically relevant biopharmaceutical property. My tutorial will be implemented through a Jupyter notebook which highlights the convenience and low-code capabilities of DeepChem for the scientific community. To demonstrate this, I will implement a few under-the-hood features to process available PROTAC datasets. My tutorial notebook can easily load data and build upon available models in DeepChem to develop a multimodal model for downstream inference. I hope this project provides a strong foundational understanding for researchers and the scientific community to explore computational methods for PROTAC design and optimization, possibly contributing to more effective research and development with downstream clinical benefits. Overall, I believe this project will contribute to the overall mission of DeepChem by democratizing computational life science research for all.
Ensembl VEP's web interface offers dozens of configuration options for variant annotation, overwhelming new users and generating recurring helpdesk queries. Ask VEPai is a locally-hosted, open-source AI assistant that translates natural-language descriptions into recommended VEP configurations with justifications. The system uses Retrieval-Augmented Generation (RAG) grounded in a curated knowledge base, with a deterministic constraint checker that catches species violations and option conflicts the model misses. Beyond the project brief's scope of option labelling, training data, and a prototype model, this proposal adds four components: (1) the constraint checker, empirically validated across three model sizes, (2) a quantitative evaluation framework with leave-one-out methodology and multi-run statistical replication, (3) a structured JSON schema mapping recommendations directly to VEP web form sections for "click to apply" integration, and (4) an interpretability layer where every recommendation includes a source citation traceable to the knowledge base. A working demo validated across Qwen 2.5 3B/7B/14B shows +19–30% Enable F1 from the knowledge base, with all code and results published. Core deliverables: expanded KB (~55 options), 15–20 gold-standard examples, RAG pipeline, constraint checker, JSON schema, evaluation report, and documentation. My availability (~400 hours) supports extending to a 350-hour Large project, adding QLoRA fine-tuning, attribution testing, FastAPI wrapper, and VEP output explainer.
Genome-scale metabolic network reconstructions (GENREs) are comprehensive maps of all known metabolic reactions and associated genes in a given organism. Starting from a genome, these reconstructions are developed using annotations, biochemical data, and information from scientific literature. This can make GENREs strain-specific. From these networks, genome-scale metabolic models (GEMs) are derived, serving as powerful tools for exploring bacterial systems. Biologically grounded constraints can be applied to these models, resulting in constraint-based models that can predict the flow of metabolites through a defined network with techniques such as Flux Balance Analysis, Flux Variability Analysis and Flux Sampling. So far, we have implemented several sampling approaches in our dingo Python library. dingo-stats is a new Python library for the statistical analysis of flux sampling datasets. It works given any flux sampling dataset (in a desired format) and thus is not restricted only to dingo samples. Functions from dingo-stats are used for identifying and removing loopy reactions (as a quality control step) and estimating their effect on thermodynamic feasibility of the samples. Additionally users can extract pathway information from the KEGG database and compare different sampling datasets with differential flux and pathways enrichment analyses. Moreover, users can calculate pairwise linear correlations and non-linear dependencies using copulas between reaction flux distributions, perform clustering techniques to predict pathways/operons and evaluate their results with several clustering quality metrics and dimensionality reduction plots. Users can also create correlation-based graphs and proceed to analysis with measuring node (reaction) centrality metrics. With these metrics, information on reaction essentialities can be inferred.
The 3D PDF Export for FreeCAD project aims to develop a native capability for exporting interactive 3D models directly to PDF format, enhancing FreeCAD's documentation and collaboration capabilities. Currently, FreeCAD users must rely on complex multi-step workflows involving external tools like MeshLab, LaTeX, or even commercial CAD software to create 3D PDFs, often resulting in loss of model fidelity, colors, and structure. This project proposes implementing a seamless 3D PDF export feature integrated directly into FreeCAD's TechDraw workbench. Users will be able to create standard technical drawings with traditional orthographic views and dimensions, plus an interactive 3D view that appears as a regular isometric view in print but becomes fully interactive when viewed digitally. Key components of the implementation include: 1. Document Structure Access: Utilizing FreeCAD's Python API to traverse the document object hierarchy or need to rely on Qt for accessing certain structural information. 2. 3D Format Conversion: Implementing exporters for both U3D format and PRC format to support NURBS surfaces for higher fidelity representation. 3. PDF Generation: Creating a PDF container that properly embeds the 3D content with interactive viewing capabilities. 4. TechDraw Integration: Extending the TechDraw workbench to include a new "3D View" type alongside traditional orthographic views. The completed feature will significantly enhance FreeCAD's usability for professional documentation, enabling users to share interactive 3D models with clients and collaborators who don't have CAD software, addressing a long-requested feature within the FreeCAD community.
Mifos Payment Hub EE is a payment orchestration engine that coordinates and routes payment transactions among core banking systems, switches, wallets, and external financial partners. The Payment hub exposes multiple APIs for payment initiation, status queries, partner integrations, and operational management, creating a large surface area for critical risks such as Broken or inconsistent authorisation across endpoints, replay or duplicate payment requests. This project will focus on conducting a comprehensive API security assessment of the Payment Hub API, with emphasis on the OWASP API Security Top 10 and real-world payment threat models. The goal is to identify, document, and help mitigate security vulnerabilities within the system’s RESTful APIs. This project will be centred mainly around: - API Surface Mapping, which deals with enumerating the APIs found in the PaymentHub and identifying the authentication flows, request headers and partner-specific parameters used. - Threat modelling deals with mapping the threats found to the OWASP category - API Penetration Testing focused on the OWASP API Security Top 10 - Tooling and Automation, dealing with developing custom scripts and reusable test cases - Reporting and Remediation, dealing with producing a detailed vulnerability report and providing fixes and mitigation to the different vulnerabilities found. By the end of the project, the following deliverables are expected: 1) A documented Payment Hub EE API threat model and API Penetration testing checklist 2) An automated or semi-automated API security test suite for reusing in other projects 3) A detailed vulnerability assessment report with remediation guidance will be provided to help developers make patches when needed and follow secure API Practices.
AIDRIN (AI Data Readiness INspector) is an open-source tool designed to assess the quality, fairness, and readiness of datasets before they are used in AI workflows. However, its current implementation supports only basic evaluations for tabular data and lacks both privacy-aware assessments and the ability to handle unstructured data formats like images. This project proposes to address these limitations in two structured phases. In Phase 1, I will enhance AIDRIN’s capabilities for tabular datasets by implementing essential privacy-preserving metrics—such as k-Anonymity, l-Diversity, t-Closeness, MM Risk Score, and Entropy-Based Disclosure Risk—alongside other key checks for class imbalance, feature correlation, and model evaluation. These assessments will be supported by intuitive visualizations to help users make informed decisions about data quality and compliance. In Phase 2, I will extend AIDRIN to support image datasets by integrating pretrained vision models to generate semantic embeddings. These embeddings will enable readiness evaluations such as duplicate and outlier detection, class imbalance analysis, completeness checks, and fairness diagnostics (when label or attribute metadata is available). The goal is to replicate AIDRIN’s structured readiness evaluation pipeline for visual data in a modular and reusable manner. Deliverables include: Privacy and imbalance metric suite for tabular data with visual summaries Embedding-based readiness evaluation pipeline for image datasets Integration with pretrained vision models (ViT, CLIP, DINOv2) CLI tool for executing and exporting assessment results Documentation, testing, and demonstration notebooks By the end of the program, AIDRIN will support robust and privacy-aware readiness assessments for both structured and unstructured datasets, helping researchers and practitioners build more transparent, reliable, and ethical AI systems.