Computer Science · Machine Learning Enthusiast
Machine Learning · Systems · Arizona State University
01 / About
I'm a Computer Science student at Arizona State University with a 4.0 GPA, focused on building machine learning systems from first principles. I care deeply about understanding architectures beyond surface-level implementation — from attention mechanisms to optimization dynamics. My goal is to design intelligent, scalable systems that create meaningful real-world impact.
Currently pursuing a B.S. in Computer Science at Arizona State University. Open to internships, research roles, and collaborative projects.
Current Focus
02 / Projects
A selection of projects spanning ML, systems, and web. Add more in lib/data.ts.
Implemented a multi-layer decoder-only transformer in PyTorch from first principles, including embeddings, multi-head self-attention, feedforward blocks, and layer normalization. Built a full training pipeline with batching, optimization, evaluation, and autoregressive text generation. Conducting experiments on training dynamics and generalization behavior.
Developed an interactive web application for real-time object detection on images and videos using YOLOv8 and OpenCV. Includes adjustable confidence thresholds, model size selection, and visual comparison tools for experimentation and performance analysis.
Built an LSTM-based time-series model predicting next-day stock direction using nine engineered technical indicators (RSI, MACD, Bollinger Bands). Implemented a custom loss function optimizing for trading profitability and applied time-series cross-validation, achieving 55% directional accuracy and 1.28% MAE.
03 / Achievements
Academic recognition, competitions, and certifications.
Maintained a 4.0 (4.2 internal) GPA during first semester in Computer Science.
Fall 2025
School Rank 1 · Zonal Rank 20 (Maharashtra & Goa) · International Rank 20.
2017
Achieved top performance in national-level computer science competition.
2014-15
Qualified for the ICPC North America Regional contest, placing in the top 20% among university teams.
Nov 2024
Modeled probability of default on a loan portfolio and applied dynamic programming techniques to categorize FICO scores for predictive modeling.
Dec 2025
04 / Skills
Languages
Machine Learning
Tools & Systems
05 / Contact
I'm open to internships, research collaborations, and interesting side projects. The best way to reach me is by email — I try to respond within a day.
ryanpateluniv@gmail.comQuick Form