Open source work by Vedant Kulkarni.
Everything I have created, maintained, and contributed to, grouped by category. Each entry notes which of the three it is.
My Google Summer of Code 2025 project with Google DeepMind, exploring, extending, and documenting the function calling capabilities of the Gemma model family. Over 12 weeks, the project grew from simple experiments to a benchmarking suite and leaderboard covering 25 Gemma variants and quantizations, a function calling SDK, a command-line interface not unlike the Gemini CLI, and a set of tutorials for developers working with offline models.
Created a custom model using the framework to simulate movement of predator and prey on a map of Central Park, NY, and added a feature that allows for easier declaration of substeps using decorators. Worked on improving the Python API, creating interactive simulations for use cases such as solar panel adoption and zoning policy, and integrating it with Beckn.
A framework that empowers Ollama LLM agents in A2A (agent-to-agent) networks with function calling.
A GPU benchmark tool for evaluating GPUs and CPUs on mixed operational intensity kernels. Modified the build process so that it auto-detects and builds the executable for the installed CUDA version, and updated the code to compile and run with CUDA 13.0.
Rewrote the library in Typescript, created the modern store interface and continue to maintain it alongside Nathan Friedly, the original author. Help in maintaining 6 out of the 8 stores written for the library, and helped rewrite the default memory store to make it more memory-efficient.
Helped with the rewrite in Typescript, and achieving feature parity with express-rate-limit. I continue to maintain it alongside Nathan Friedly, the original author.
Refactored the library, added tests, and transitioned to ESM + Typescript.
Yet another library to extract text from docx, pptx, xlsx, and pdf files. What sets it apart is that it parses files based on their mime types, and not their file extensions. While the parsers for pdf, docx, and xlsx files use popular libraries, the parser for pptx files is written and maintained by me.
Fixed a bug where the library does not export-default an expression when re-exporting the default export from another file.
A HTTP content negotiator for V that allows you to parse the Accept header of an HTTP request and get the preferred response media type specified by the client.
Updated the library to work with the latest version of V.
An implementation of JSON Web Tokens for V that creates and verifies tokens signed with HMAC-SHA224, SHA256, SHA384, and SHA512.
Contributed to 8 popular features, including useful-not-found-page, restore-file, quick-comment-edit, linkify-code, and forked-to.
Restructured the documentation, rewrote parts of the installation page, added a getting started page, tutorials, and examples, as well as a CLI flags reference page.
Fixed a bug where the rendered diagram is printed twice in the terminal.
Added a host validation check to the flutter doctor tool.
A minimalistic pacman wrapper written in bash.
A minimalistic backup tool written in bash.
Created a script to automate adding plugins to the official channel.
A plugin that formats your code using Prettier for my favourite editor, Micro.
A plugin that lints your code using XO for my favourite editor, Micro.
Documented the process of creating and working with registries, and created the registry-cli package to spin up a registry quickly and easily. Submitted various RFCs to improve the project.
Helped create a site for developer documentation using Gatsby and Astro, and dockerized the reference Beckn app implementation.
The specification and minimal reference implementation for a plugin-based service that allows issuers to render verifiable presentations, based on the W3C verifiable credential specification from templates, and store it in a queryable database for holders to list/retrieve.
Students at IIIT Hyderabad use it to register for the mess and check their bills. It automates a lot of the administration’s paperwork, such as billing. I co-created it with Aarnav Pai, and continue to maintain it and the associated systems.
The Dabbu Knowledge Platform was created with an aim to rethink the way we organize and traverse large amounts of knowledge, no matter where it is stored. Dabbu allows you to access any of your personal information (Gmail, Google Drive, OneDrive, your hard drive, etc.) as simple files and folders from the Dabbu CLI.
You can also go into the knowledge drive where you can pivot information
by topics/people/places. For example, k:/$ cd austin will return you
all your information from Gmail, GDrive, OneDrive that has a reference
to the place Austin. The feature that allows you to organise/search your
files based on topics in the files’ text uses the Latent Dirichlet
Algorithm to extract the topics.
A simple static site that allows you to explore and search all the projects in the GSoC 2025 program, using data fetched from the Summer of Code API.
A simple static site that allows you to import, export, and print your friend group’s joint timetable.
tiny, adorable urls: a url shortener, written in python using flask and sqlite3.
A python tool that uses Deeplator to translate subtitles in .srt files to and from six different languages. This project was one of my first open source projects, and my submission to Google Code-In in 2018.
I participated in the Hackathon with my younger brother to reimagine how kirana store owners would update their catalog online. My brother did the UX designs, while I helped create a Beckn-powered Svelte web component that enabled voice based search and creation of items in a crowdsourced product catalog. Our submission won a special mention award from the jury.