Explore Projects

Concerto, the core modeling language of the ACCORD PROJECT, is critical for smart legal contracts and domain modeling. This project aims to migrate from the current testing framework mocha to vitest which will enhance the performance, execution time and debugging capabilities. This process can be automated by the combination of vitest-codemod, regex transformations, ai assistance and some manual scripts required for handling complex test cases which cannot be automated. Here are the deliverables: 1. Migration from Mocha, Chai, Sinon to Vitest for all Concerto implementation. 2. A new Concerto conformance package which will be a standalone validation test package for all Concerto implementations.

Making plugins for improving the quality of the dataset to improve model performance. Various standard techniques in hyperparameter tuning are to be used to get the ideal dataset configuration.

<p>Add networks/orchestrator to run distributed dataflows.Add new tutorials on using DataFlow/Operations and how models/database abstractions etc interoperate within a flow.</p>

I will be adding new UI elements and improving the existing UI system to easily visualize and transform various geometric shapes and UI elements interactively without affecting the existing performance.

<p>The project highlights use-cases that explain the Machine Learning workflow of DFFML API, through Jupyter notebook examples. The project also adds multi-output model support to DFFML.</p>

This proposal introduces an optional, privacy-first AI assistant for OpenSCAD to flatten the learning curve of SCAD syntax and geometric transformations. Integrated as a togglable dock panel, the assistant remains completely inert when closed and requires no cloud subscriptions or external API keys, instead leveraging locally running models via Ollama or any OpenAI-compatible server. Beyond standard Q&A, the project implements agentic behavior where the AI can analyze the active script, propose modifications through an inline diff, and apply approved changes directly to the editor as undoable actions that trigger an automatic preview. The architecture is designed to be vision-model-ready from the start, ensuring future viewport context integration requires no structural refactoring. This creates a transparent, local-only loop that empowers users without compromising the project’s core philosophy of user-controlled development.

This project aims to enhance the performance of Generative AI (GenAI) workloads on ARM devices using the OpenVINO toolkit, focusing on reducing latency and throughput metrics, accelerating model compilation times, and minimizing memory requirements. The initial phase involves establishing a robust benchmarking environment tailored for GenAI models, employing tools such as Flame Graphs for performance visualization and monitoring tools for memory usage analysis. This setup will enable rapid assessment of optimizations' impacts on performance and resource consumption. Refine and enhance the performance of Generative AI workloads on ARM devices, targeting the OpenVINO toolkit. The project will concentrate on advanced optimization techniques not yet fully explored or implemented in the OpenVINO ecosystem for ARM architectures, specifically focusing on graph transformations, low-level kernel optimizations, and enhanced utilization of the ACL (Arm Compute Library).

This project introduces a ‘Cold Start’ Data Factory that turns zero/one-shot prompting into a full data-to-training workflow. The core contribution is a production-grade DatasetWriter pipeline that converts inference outputs into validated, reproducible annotations, supports format-aware serialization, and guarantees reliable flushing/finalization for large export jobs. The generated datasets can be exported in standard formats such as YOLO, COCO, and CVAT, with quality statistics and filtering controls for practical downstream use. Building on that, the project adds direct OTX training integration: once export is finalized, the system can automatically generate training configs, spawn OTX runs, stream logs/status, and track artifacts and run metadata. This shifts geti-instant-learn from a live-inference-only experience to a complete cold-start bootstrapping system where prompt-based “Teacher” labels are transformed into trainable “Student” datasets and model training jobs. The result is a faster and more reliable path from a few prompts to deployable edge models.

<p>Currently , Tardis has the capability to store the state of model to an HDF5 file. This project , is mainly concerned , to be able to restore that Simulation Object , back from HDF5 file , which is identical to the object used to produce the file. Also, It should be possible to analyse or continue simulation on the restored Simulation Object.</p>

This project aims to implement a methodological approach to further improve the visualization of DTI tensors and HARDI ODFs glyphs by using well-established techniques in the field of computer graphics such as billboarding, signed distance functions (SDFs), and ray marching. In addition, it is proposed to visualize the uncertainty associated with the DTI model.

I will be working on integrating the glTF file format to FURY which shall enable us to load, visualize, and animate a glTF model. I will also add unit tests & demos for the newly integrated file format. As part of my stretch goals, I'll be working on integrating the GLB format which is a compressed version of glTF.

<p>Among the 3D formats available, some are simple in theory, yet effective for a lot of different use cases and supported by a multitude of software in the industry. The challenge they offer is the number of iterables. Stanford PLY, for example, quickly gets over a million vertices. STL being a lossy format, has to be stored with extra details, making it enormous. Importing such models faster and doing so in the memory limits is the aim of this project.</p>

<p>The bevel modifier is extremely powerful, but there is constantly a list of requested improvements that could expand its use case and speed up the modeling process. One of these is user-drawn profiles, which is a commonly requested feature that is somewhat separate from the main functionality of the bevel operation, which makes it a good candidate for a GSoC project. There has been successful GSoC projects with the Bevel modifier in the past few years, and I hope with this project I can continue that success.</p>

The Pharo VM project is a continuation of the opensmalltalk-vm (or Cog VM) project which is the virtual machine underneath Pharo, this project will be centered around the Pharo-to-C transpiler, also called Slang. In this project my primary target is to improve the development experience for the Pharo VM and to accomplish this I will be implementing: Type feedback improvements, type guided translations, incremental building of the AST and integration of the C model to the IDE.

PyElastica is the Python version of Elastica, a project that uses Cosserat Rod theory to simulate assemblies of thin one-dimensional bodies. Currently, in PyElastica we can model the contact of rods with a frictional plane or rigid cylinder and themselves. The following are the deliverables : 1.) Develop an algorithm to import STL mesh and convert it into PyElastica Geometry. 2.) Develop a library that implements contact forces between the rod and the imported shape. 3.) Validate implemented features/craft examples.

<p>This proposal aims to bring a selection of much needed improvements to the knife tool in Blender which will better the modelling workflow of the tool. It proposes such changes as the ability to undo individual steps of the cutting process, a visual measure of distance and angles per cut, more precise angle control in constrained angle mode, the ability to snap cuts to global or local orientation, knife usage in multi-object edit mode and an edge removal mode.</p>

<p>The bevel modifier is widely used in modelling to chamfer edges and smoothen them out. Bevels are used to correct the shading of mesh, so as to make the edges blunt to give a solid and realistic look as opposed to non beveled objects. However the current modifier does not ensure correct normals and has various listed bugs which slows down artists workflow. Having written my own Weighted Normals modifier in last years Google Summer of Code. I am familiar with the modifier stack and its implementation and can work to provide good quality functionality on the modifier and tool.</p>

<p>I will introduce methods to pgmpy to select Bayesian models based on data sets. First, I will implement support for basic score-based and constraint-based structure learning. Second, I will add common enhancements to the score-based approach, including local score computation + memoization and tabu lists. Finally, I will implement the MMHC algorithm, which combines the score-based and the constraint-based method.</p> <p>I believe that structure learning is <em>the</em> missing feature in pgmpy at the moment.</p>

<p><strong>Abstract</strong>: Scrapy is one of the most popular web crawling and web scraping framework. It’s written in Python and known by its good performance, simplicity, and powerful API. However, it’s only possible to write Scrapy’s Spiders using the Python Language. The goal of this project is to provide an interface that allows developers to write spiders using any programming language, using json objects to make requests, parse web contents, get data, and more. Also, a helper library will be available for Java, JS, and R.</p>

<p>Data structures that complex networks (graphs) can model are present almost everywhere. Therefore, it is important to have a software that can visualize and provide insights into such structures. Unfortunately, the vast majority of network visualization and manipulation software currently available has several limitations. For example, Gephi (Java) does not scale well when the number of nodes increases and cannot deal with 3d layout algorithms. Because of those limitations in the currently available network visualization software, we propose to develop a client/server network visualization for FURY. Our proposed system is inspired by the following work: “Harnessing WebGL and WebSockets for a Web-Based Collaborative Graph Exploration Tool”.</p>

<p>The Activeloop organization provides data scientists worldwide with a solution to these problems so that they can spend more time training their models and saving resources. Hub is a package that stores petabyte-scale datasets as a single numpy-like array on the cloud, so one can seamlessly access and work with it from any machine. Hub has a simple API that enables its users to obtain a lot of functionality with just a few lines of code. However, as a complete beginner, it is difficult for many to learn the nuances of a new package. Reading the documentation is something that helps, but it is not exactly a fun way of learning something new. It can become tedious and boring. This is exactly why it would be great if Hub had a package that interactively teaches the basics of hub in the CLI itself.</p>

<p>Dimension reduction of a molecular dynamics (MD) simulation is meant to provide insight into the slow underlying dynamical motions of a set of molecules. Dimension reduction algorithms are either nonlinear or linear transformations, both categories providing distinct benefits and shortcomings. Existing implementations of any dimension reduction algorithm work by decomposing either some subset of an N-by-P “observation matrix” delineating the N “observations” of a system with P coordinates or with a P-by-P correlation matrix. Functions in MDAnalysis avoid creating such a matrix as it would proscribe users from analyzing very large MD simulations on a workstation or laptop. The goal of this project will be to implement Principal Component Analysis and diffusion maps in MDAnalysis such that both methods are accessible to aforementioned users.</p>

PyZombis is a community course designed to teach Python to the Spanish-speaking community, based on the Brazilian MOOC "Python para Zumbis." The idea is to create an improved Spanish version with an interactive environment for students, using the Runestone library to visualize and try active code. Currently, PyZombis uses the Brython Interpreter, which lacks support for modern libraries like Pandas and Numpy. To overcome this, I propose integrating PyScript, enabling rich Python applications within the browser and providing access to more libraries. Additionally, PyZombis requires database functions for authentication and tracking user progress. I plan to create a SCORM (Sharable Content Object Reference Model) compliant package and use it with an LMS like Moodle to monitor and track progress by integrating it with PyZombis.