"We do what we must because we can." - GLaDOS
📊 AI/ML Engineer at Perfect Solutions
💡 Focus: Designing, training, and deploying machine learning models for enterprise applications
🔍 Responsibilities:
- Building PyTorch and TensorFlow models for automation and AI-driven use cases.
- Developing data preprocessing pipelines for multi-million-row datasets.
- Supporting end-to-end ML workflows, from data to deployment.
📖 Learning:
- Rust – Exploring the basics of ownership and how to use in my ML cycles
- Deep-ML – Solving Machine Learning and Data Science problems
- Improving my LeetCode Problem-Solving skills for interviews
💡 Working On:
- RAG for Vintage Keyboards & Switches – Building a retrieval-augmented generation system using FAISS & FastAPI to preserve knowledge from the now-defunct Deskthority Wiki
- Vintage Board Modernization – Tinkering with classic keyboards by creating custom keymaps from scratch and adapting them for modern firmware (QMK/ZMK)
- Improving My GitHub Repos – Making code more production-ready
- Building My Website – Hosting my portfolio & projects
🎓 MS in Machine Learning, Stevens Institute of Technology
🖥️ Artificial Intelligence Enthusiast
💡 Interests: Computer Vision, NLP, Generative AI, MLOps, LLMs
🎮 Fan of the Portal Series
🔹 Spotify Insights Lab – Full-stack ML dashboard analyzing 1.2M+ tracks, with XGBoost promotion model, Streamlit UI, and PostgreSQL query optimization (~150× faster).
🔹 Helmet Detection with YOLOv5 – Custom YOLOv5 model trained from scratch on 5K images; evaluated with FiftyOne to debug predictions and performance.
Here’s my latest contribution:
QMK Firmware - ANSI Tsangan Layout for Ducky One 2 Mini
