Jeevan
AI & Software Developer | Creator of AgentForge QA | Kornia Open-Source Contributor | AI Agents & Robotics
Hyderabad, India
I'm Jeevan Kumar, a second-year Computer Science student specializing in Artificial Intelligence and Data Science, based in Hyderabad. I build AI-powered systems, developer tools and automation projects, with a strong focus on software reliability, intelligent agents and robotics. I'm the creator of AgentForge QA, an open-source CLI that helps developers verify AI coding-agent work using real repository evidence. It analyzes test results, Git state, required artifacts and development claims, helping identify incomplete or potentially risky changes. I also contribute to Kornia, a widely used open-source computer vision library built on PyTorch. My contributions focus on bug fixes, numerical correctness and regression testing. I'm currently developing a robotic chess system using ROS 2, Gazebo and a Franka FR3 robotic arm, exploring robotic manipulation, simulation and intelligent decision-making. I enjoy building products from initial research and architecture through implementation, testing and deployment. My primary interests include AI agents, trustworthy AI, developer infrastructure, computer vision, open-source engineering and robotics. I'm always interested in connecting with developers, researchers and teams building meaningful technology.
Open-Source Contributor
Kornia
- Contributed bug fixes and regression tests to Kornia, an open-source computer vision library built on PyTorch.
- Investigated correctness issues involving geometry, image transformations and numerical edge cases.
- Created focused fixes supported by reproducible test cases.
- Collaborated with maintainers through GitHub pull requests, code reviews and continuous integration checks.
Independent Software Developer
Independent Projects
- Created AgentForge QA, an open-source CLI for verifying AI coding-agent work against deterministic repository evidence.
- Built verification workflows covering tests, Git state, required artifacts, development claims and risk assessment.
- Implemented a failure-suppression scanner to identify suspicious error-handling patterns in changed code.
- Managed product research, architecture, development, automated testing, documentation and release workflows.
An open-source, local-first CLI that verifies AI coding-agent work using deterministic repository evidence. Checks development claims against configured tests, Git state and required artifacts, producing structured reports and conservative verdicts. Includes a failure-suppression scanner to flag suspicious error-handling patterns for human review.
Open-Source Software Engineer (Contributor)
Contributed to Kornia, a widely used open-source computer vision library built on PyTorch, improving its documentation, API reliability, numerical correctness and test coverage. Expanded Python docstring coverage across major repository modules, including augmentation, image processing, geometry, losses, tracking and utilities, documenting public APIs, parameters, return values and usage examples. Progressed into resolving functional bugs and numerical edge cases involving geometric transformations, homography, epipolar geometry, image augmentation, tensor shapes and batch handling. Developed targeted regression tests to reproduce reported failures, validate fixes and guard against future regressions across different input shapes, data types and execution conditions. Collaborated with upstream maintainers through GitHub issues, pull requests, code reviews and CI validation, delivering focused changes that improve the reliability and maintainability of the library.
Robotic Chess System — Franka FR3
An ongoing robotics project developing an intelligent chess-playing system around a Franka FR3 robotic arm. Uses ROS 2 and Gazebo to investigate robotic control, gripper actuation, simulation physics, chess planning and safe execution. Currently focused on validating reliable manipulation before hardware integration.
An ongoing humanoid robotics initiative exploring the architecture of intelligent robotic systems. Focused on modular foundations for perception, decision-making, control, simulation and future autonomous capabilities. The project combines robotics research, system design and iterative software development.
Gitomon
Gitomon — Next-Generation Developer Experience Platform An independently developed, full-stack platform exploring a new approach to developer engagement and personalized digital experiences. Engineered secure GitHub integrations, deterministic backend systems, persistent data management, and interactive application workflows. Designed with a strong emphasis on privacy, scalability, user experience, and extensible architecture. Built using React, TypeScript, Node.js, Express, PostgreSQL, and Drizzle ORM. Status: Private Product | Pre-Launch Development
B.Tech — Computer Science and Engineering (AI & ML)
NxtWave Institute of Advanced Technologies (NIAT) · 9.60-currently sem's-grade
Machine Learning Pipelines with Azure ML Studio
Coursera
Feb 20265-Day AI Agents Intensive Course with Google
Kaggle
Dec 2025
Attendee · Oct 24, 2026