GSoC 2026 Interest: Building on my previous contributions (PR #175990)

Hi PyTorch Team,

I am M Husnain, a BSCS student at MNS UET Multan with a 3.8 CGPA. I have a deep-rooted interest in Neural Networks and have completed several specializations including Deep Learning, TensorFlow Developer, and PyTorch for Deep Learning.

I recently got my PR #175990 approved, which addressed numerical inconsistencies in Conv3d. This experience gave me a good grasp of PyTorch’s testing and regression frameworks.

For GSoC 2026, I am eager to contribute further. Beyond my PR work, I am architecting ‘Sentinel-Agent’, an autonomous multi-agent framework using LangGraph, LLMs (Llama 3/GPT-4o), and Docker for real-time code vulnerability detection.

Specifically, I am interested in Improving Performance of Deep Learning Kernels. I look forward to guidance from mentors on how to refine my proposal.

Best regards, M Husnain

Reference PR: Fix numerical inconsistency in Conv3d.reset_parameters for channels_l… by mhusnain-tech · Pull Request #175990 · pytorch/pytorch · GitHub

Project Proposal (Sentinel-Agent): https://drive.google.com/file/d/1e5fWt2xzHCGpO9wd1lhuOVSGxE0NwNIG/view?usp=sharing