Data Scientist | AI Researcher | Full-Stack Developer
Building AI-powered applications & intelligent systems

IntelliTrainer – AI-Powered Exercise Tracker
Developed under the AI and ML at SJSU organization. The official project page is hosted at
https://www.ai-ml-club-sjsu.com
but it is currently private since it is owned by the organization.
Pose-driven fitness assistant focused on precise, per-exercise form evaluation rather than generic exercise classification.

• Rebuilt the pipeline after midsemester testing to remove unnecessary exercise detection and instead apply targeted, rule-based analysis for each specific movement.
• Improved joint tracking stability, angle consistency, and rep detection through refined MediaPipe processing and cleaner state logic.
• Designed the system for modular growth, including future support for LLM-based coaching, user progress modeling, and GPU-accelerated inference.

PaperLinks – AI-Powered Research Digest Generator (developed @ CalHacks 12.0)
View on Devpost
Built at CalHacks 12.0, PaperLinks is an AI-driven research platform that turns a single topic prompt into concise, clustered digests of recent arXiv papers.

• Backend built with FastAPI integrates MiniLM embeddings and Chroma Cloud for semantic clustering and storage.
• Uses Claude 3.5 Sonnet via Lava proxy to generate summaries, topic clusters, and executive overviews.
• Frontend built with Next.js 14 + Tailwind, featuring a dynamic “Browse” and “Digest” mode for visual exploration.
• Added Fish Audio TTS support for on-the-fly voice playback of research summaries.
• Caches and reuses digests for fast reloads and offline support.
• Designed and deployed during a 36-hour hackathon for real-time arXiv paper discovery and AI-powered summarization.

Predictive Analytics for Safer Roads
Traffic accident severity prediction system using PyTorch, Plotly Dash, and Google BigQuery.

• Built neural network + statistical models on a dataset of 7.7M accidents.
• Developed an interactive dashboard deployed on Google App Engine.
• Conducted advanced statistical tests and feature engineering with weather & road data.

Club Hub – University Club Management Platform
Full-stack web application for student organizations using React, Node.js, MySQL, and Express.

• Implemented user registration, event tracking, and club workflows.
• Designed scalable backend architecture and interactive UI.

Food Image Classification
Deep learning image classifier using Inception V3, VGG16, and Keras/TensorFlow.

• Applied transfer learning with advanced preprocessing (data augmentation & normalization).
• Achieved high accuracy on multi-class food dataset.

Expanding into large-scale AI systems, multimodal ML, and applied research, with a focus on building tools that improve real-world workflows.

• Leading a 10+ member IntelliTrainer project under AI and ML at SJSU, developing an intelligent exercise tracking platform with rule-based and ML pipelines, GPU acceleration, and future integration with LLM-based coaching.
• Increasing participation in university and national hackathons to explore new ideas, build prototypes quickly, and continue improving through competitive development environments.
• Learning advanced concepts in AI agents and autonomous workflows, including planning, tool use, long-context reasoning, and memory-based agent architectures.
• Developing an AI agent system to automate key parts of college life, such as class scheduling, assignment tracking, email drafting, study planning, and productivity analysis.