Tools

Open Source Tools

Standalone tools and utilities I've built. Each lives in its own repo and is deployed independently.

PolyPulse

Live

Paper-trading engine with live sentiment analysis

A real-time multi-source signal aggregation and paper-trading engine. Combines live market tape data, news sentiment analysis, and TradingView webhook signals to score assets and execute simulated trades with adaptive risk management.

  • Live market data from Yahoo Finance (crypto & forex)
  • Multi-source sentiment scoring
  • Three bankroll simulation modes
  • Real-time SSE streaming dashboard
  • TradingView webhook integration
  • Configurable strategy engine
Node.jsTradingSSESentiment Analysis

Markdown Viewer

Live

Feature-rich markdown viewer and editor with desktop downloads

A full-featured markdown viewer and editor supporting GFM, LaTeX math, Mermaid diagrams, syntax highlighting for 190+ languages, and a live split-pane editor. Available as a web app and as a native desktop app for macOS, Windows, and Linux.

  • Full GFM support with tables, task lists, and footnotes
  • LaTeX math rendering via KaTeX
  • Mermaid diagram support (flowcharts, sequence, state)
  • Syntax highlighting for 190+ languages
  • Live split-pane CodeMirror 6 editor
  • Desktop downloads for macOS, Windows, and Linux
ReactTypeScriptElectronMarkdownVite

LLM API Playground

Live

Multi-provider LLM gateway with chat UI

A FastAPI-based LLM gateway supporting OpenAI GPT-4o Mini, Anthropic Claude Sonnet, and Google Gemini Flash. Features text summarization, sentiment analysis, interactive chat with conversation history, and prompt engineering with 3 strategy variants.

  • Multi-provider LLM support (GPT-4o, Claude, Gemini)
  • 3 prompt engineering strategies per endpoint
  • Interactive chat with conversation history
  • Sentiment analysis with confidence scoring
  • Text summarization with configurable length
  • Prompt enhancement using engineering best practices
PythonFastAPIOpenAIAnthropicGemini

RAG Document Q&A

Live

Interactive RAG pipeline over RL research papers

A full RAG system built over 10 reinforcement learning research papers. Features an interactive 6-step notebook, live Q&A chat with hybrid BM25 + vector search, dual evaluation (heuristic + LLM-as-judge), and one-click autoplay that runs the entire pipeline.

  • Interactive 6-step RAG notebook
  • Hybrid BM25 + vector search retrieval
  • Live Q&A chat grounded in research papers
  • Dual evaluation: heuristic + LLM-as-judge
  • One-click autoplay for full pipeline execution
  • Downloadable results, charts, and documentation
PythonFastAPILangChainOllamaChromaDBRAG