Skip to content
View py-xis's full-sized avatar

Highlights

  • Pro

Block or report py-xis

Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
py-xis/README.md

py-xis

I'm a Computer Science student at the International Institute of Information Technology, Bangalore, drawn to the space where systems engineering meets machine learning. I enjoy building distributed systems and deployment infrastructure as much as I enjoy training and fine-tuning generative models — and I'm happiest on projects that let me do a bit of both.


💻 Tech Stack:


🛠 Projects

🔧 Related to Software Development

  • GoPloy
    A self-serve deployment platform for React apps — clones, builds, and serves a live site behind a custom subdomain from just a GitHub URL, with real-time build-log streaming.
    Built with: Go, Docker, WebSockets

  • ReSolve
    A platform to track unsolved problems from Codeforces, LeetCode, and AtCoder to assist in interview prep.
    Built with: Node.js, Express, React, Tailwind CSS

  • Ditto
    A cross-platform shared clipboard app that enables seamless encrypted text synchronization across devices.
    Built with: Python, Firebase

  • ReadSphere
    CLI-based online library management system using client-server socket programming with full CRUD support.
    Built with: C, Threads, IPC

  • CampusManager
    Menu-driven interactive school database manager handling students, teachers, and courses via JDBC.
    Built with: Java, MySQL

☁️ MLOps / DevOps

  • FlowGuard
    An end-to-end MLOps pipeline for a spam-classification model — Jenkins-triggered CI/CD into a self-healing Kubernetes deployment, with Prometheus metrics and ELK-based logging for real-time observability.
    Built with: Python, Docker, Kubernetes, Jenkins, Prometheus, ELK Stack

🧠 Machine Learning / Computer Vision

  • VisCommerce
    A Visual Question Answering system for e-commerce — curated a 93,000-pair dataset from Amazon product listings and fine-tuned BLIP and PaliGemma with LoRA, reaching 76.3% accuracy and 0.89 semantic similarity.
    Built with: Python, PyTorch, LoRA

Pinned Loading

  1. py-xis py-xis Public