Projects

A portfolio of quantitative trading systems, machine learning models, and fullstack applications that demonstrate professional-grade engineering and measurable impact.

Independent Projects

Breakout Study Tool

Next.js, TypeScript, React 19, Express/tRPC, FastAPI, Supabase, Prisma, Turborepo

View Live Site

Built a production web app for studying breakout patterns across ~10,000 tickers using yfinance and pandas, with a Next.js + TypeScript frontend. Features subscription-based monetization with free and premium tiers. Currently expanding with a live AI Insight System for automated setup detection, performance analytics, and adaptive signal scoring.

Key Highlights

  • ▸Scans multiple timeframes (1/5/15 min, 1h, daily, weekly) with dynamic chart overlays, tagging, and up/down pattern classification
  • ▸Processes ~10,000 tickers using yfinance and pandas
  • ▸Subscription-based monetization with free and premium tiers for user access
  • ▸Integrated Supabase (Auth + storage) and deployed on DigitalOcean
  • ▸Used daily for real-time pattern practice and analytics by multiple users
  • ▸Currently expanding with a live AI Insight System for automated setup detection, performance analytics, and adaptive signal scoring to forecast breakout quality and trader accuracy trends
  • ▸Full-stack architecture using React 19, Express/tRPC API, Python/FastAPI data processing, and Supabase/Prisma database in a Turborepo monorepo
Next.jsTypeScriptReact 19ExpresstRPCFastAPIPythonyfinancepandasSupabasePrismaTurborepoDigitalOceanTradingAI

X (Twitter) Lead Scraper

Python, Web Scraping, Automation

View on GitHub

Developed an automated Python scraper to identify and rank potential users for the Breakout Study Tool by analyzing posts on breakout trading, technical analysis, and market trends, filtering results by keywords and engagement metrics.

Key Highlights

  • ▸Automated lead generation for Breakout Study Tool platform
  • ▸Analyzes X (Twitter) posts on breakout trading, technical analysis, and market trends
  • ▸Filters results by keywords and engagement metrics to identify potential users
  • ▸Supports platform growth through systematic lead identification
PythonWeb ScrapingAutomationLead GenerationTrading

Live Intraday Scanner

Thinkorswim + JavaScript

View on GitHub

Wrote a JavaScript pipeline that feeds custom scans into Thinkorswim watchlists for real-time candidates; notably surfaced CLSK ahead of a breakout for discretionary review.

Key Highlights

  • ▸JavaScript pipeline for real-time market scanning
  • ▸Custom scans into Thinkorswim watchlists
  • ▸Notably surfaced CLSK ahead of a breakout for discretionary review
  • ▸Real-time candidate identification for trading opportunities
JavaScriptThinkorswimTradingReal-timeMarket Scanning

Cat-vs-Dog Classifier

TensorFlow, Local GPU

View on GitHub

Collaboratively trained a CNN on a local GPU to classify images with solid accuracy; managed compute usage (batch sizing/runtime) to run reliably on limited VRAM—giving hands-on experience with GPU workflow basics.

Key Highlights

  • ▸Collaboratively trained a CNN on a local GPU
  • ▸Classified images with solid accuracy
  • ▸Managed compute usage (batch sizing/runtime) to run reliably on limited VRAM
  • ▸Hands-on experience with GPU workflow basics and resource optimization
TensorFlowCNNGPUMachine LearningComputer Vision

Academic Projects

LLM Energy Benchmark Research

Built an end-to-end pipeline for measuring energy consumption across LLM models (OpenAI GPT-4o-mini, Groq Llama-3.1-8b, Mistral Large) using CodeCarbon. Implemented streaming data collection from LMSYS Chat 1M dataset, prompt extraction with validation, and multi-API integration with exponential backoff retry logic. Engineered 30+ linguistic features using spaCy, NLTK, and textstat. Research paper currently in publication process.

PythonCodeCarbonspaCyNLTKtextstatNLPMachine LearningResearch

CS 61B Data Structures & Algorithms

Implemented ArrayDeque and LinkedListDeque (circular sentinel), BSTMap, and project features (iterators, equals, toString, resizing). Wrote JUnit tests and analyzed runtime; explored how maps/queues can model basic order-book mechanics.

JavaAlgorithmsData StructuresJUnit

Self-Directed AI/ML Coursework

Completed practical courses covering tensors & autograd, nn.Module, DataLoaders, training/evaluation loops, overfitting control (regularization/early stopping), and basic model types (MLP/CNN); used NumPy/Pandas for preprocessing and small applied exercises.

PyTorchTensorFlowNumPyPandasMachine Learning

Supplemental Online Study

Python & backend fundamentals, plus quantitative topics (stochastic processes, numerical optimization, introductory quantitative finance) with small practice projects.

PythonBackendOptimizationFinanceStochastic Processes

Let's Build Something

Open to projects, research collaborations, and innovative work in AI, finance, and technology. If you're building something that matters, I want to hear about it.

Get in TouchView GitHub