Backend engineer building autonomous agents that solve real problems.
33 agents built — from stock trading bots to code auditors, all production-ready.
| Project | What it does | Stack |
|---|---|---|
| auto-trader | Automated stock trading with 17 technical indicators + candle patterns + supply/demand analysis | Node.js, KIS API |
| invest-quant | 8-agent quant pipeline — factor scoring, walk-forward validation, drawdown monitoring | Node.js, Python, DART API |
| mcp-gateway | MCP agent hub integrating 15 tools across 6 domains | Node.js, MCP SDK |
| ghostcode-auditor | Shadow logic detector — tree-sitter AST + git blame + 15 rules + 4 metrics | Python, FastAPI, tree-sitter |
| shadow-english | YouTube subtitle-based shadowing + CEFR vocab + SM-2 spaced repetition (zero LLM) | Node.js, NLP |
| news-scraper | 8-source news aggregation + scoring + Anti-Echo-Chamber counter-perspective generation | Node.js, Cheerio |
Languages Node.js · Python · Bash
Infra Linux (WSL) · cron · MCP · REST
Data DART API · KIS API · RSS · Web Scraping
Analysis AST (tree-sitter) · Technical Indicators · Factor Models
- LLM-optional: 15 of 33 agents run with zero LLM calls — pure rule-based logic
- Security-first: 3-layer secret defense (hook → pre-commit → daily scan)
- Autonomous: Agents self-monitor, self-heal, and report via email/Notion
