QuantCoder is a command-line tool that converts academic research articles into executable QuantConnect trading algorithms. It uses a multi-agent AI architecture to extract strategy logic from papers, generate Python code, compile against the QuantConnect API, and iterate on errors automatically. Includes an interactive chat mode for strategy refinement.

Key Features:

• Research paper to trading algorithm conversion
• Multi-agent architecture (extractor, coder, compiler, fixer)
• Interactive chat for strategy development and backtesting
• QuantConnect API integration (compile, backtest, deploy)
• Support for multiple LLM providers (OpenAI, Anthropic, Ollama)

Chat with Fundamentals is a comprehensive financial research and portfolio management platform that combines AI-powered fundamental analysis with advanced quantitative portfolio optimization. Built as a research engine to explore how autonomous agents and AI can automate and enhance financial analysis, it features multi-agent AI analysis, shares-based portfolio tracking, and RAG systems for quantitative research and SEC filings.

Key Features:

• Multi-agent AI analysis (stocks, ETFs, forex, macro indicators)
• Shares-based portfolio management with 5 optimization strategies
• Comprehensive EODHD API integration (50+ endpoints across 9 categories)
• RAG systems for quant research papers and SEC filings (10-K, 10-Q, 8-K)
• Advanced risk analytics (Monte Carlo, VaR, CVaR, rolling Sharpe ratios)
• Real-time agent console with WebSocket logging

WindMar is a weather-aware maritime route optimization system designed for MR product tankers. It uses A* pathfinding over ocean grids with real meteorological data to find fuel-optimal routes that account for wind, waves, currents, and vessel physics. The system models hull resistance, propulsion efficiency, and cargo loading to produce realistic fuel consumption estimates.

Key Features:

• Weather-aware A* routing with ECMWF/GFS meteorological data
• Physics-based vessel modeling (hull resistance, propulsion, cargo effects)
• Fuel consumption minimization across real ocean conditions
• Interactive map visualization with route comparison
• Docker Compose deployment (backend + frontend + worker)

MarChat is a maritime shipboard AI assistant that provides RAG-powered search over maritime regulations (SOLAS, MARPOL, STCW), multi-mode chat for different operational contexts, and an automated pre-arrival document engine. Uses local-first LLM inference via Ollama with hybrid search combining semantic embeddings, BM25, reciprocal rank fusion, and cross-encoder reranking.

Key Features:

• Hybrid RAG search (semantic + BM25 + RRF + cross-encoder reranking)
• Multi-mode chat: regulatory, medical, general, forms
• Pre-arrival form auto-fill from natural language context
• Excel template processing with field detection
• Multi-collection document management
• Local-first inference via Ollama (offline-capable)