Built a full-stack self-hosted investment intelligence platform that monitors 150+ live market symbols and generates AI-driven trading signals, spike predictions, and portfolio recommendations in real time
Engineered a fire-and-forget job architecture using nohup detached processes and database polling to decouple long-running Ollama AI jobs from HTTP request lifecycles, bypassing nginx timeout constraints
Integrated Yahoo Finance screeners to dynamically surface market opportunities across most-actives, top gainers, growth stocks, and undervalued large-caps — no hardcoded watchlists
Built a 1-minute candle spike detection engine that identifies pre-spike conditions (Bollinger Band squeezes, volume buildups, RSI extremes, MACD crossovers) before price moves occur
Implemented a multi-category Top Picks system that generates one AI-curated recommendation per strategy type (best, safe, aggressive, options, crypto, ETF, short-term, swing) with entry zones, targets, stop-losses, and full trading strategies
Developed live price reconciliation logic to override stale AI-generated prices with Yahoo Finance's regularMarketPrice client-side, ensuring charts and card data always match
Built signal deduplication that excludes recently-analysed symbols from subsequent runs, ensuring each scan surfaces fresh opportunities rather than repeating prior picks
Designed a multi-user authentication system with session management, per-account watchlists, portfolios, and alerts, plus admin-only user management with email credential delivery via iRedMail
Created an interactive chart modal with crosshair price tracking, OHLC candlestick rendering, and hover-based price inspection across all chart surfaces
WebBetas