AI-driven quantitative stock selection system based on Qlib + LightGBM with dual LLM fallback architecture.
Key Capabilities:
- Dual LLM fallback (Claude primary + DeepSeek fallback) for high-quality stock analysis reports
- RD-Agent factor discovery auto-injected into LightGBM strategies via standardized IC validation pipelines
- “Discovery → Validation → Production” closed loop
- Combining local ML precision with cloud LLM versatility
12+ years of team management experience leading 30+ person teams, with a peak of 60+ technical staff; comprehensive experience in corporate strategy, cost control, and talent development; 80% of time spent on frontline execution;6+ years of CEO experience in Internet finance, building and operating teams from zero to one.
15+ years of full-stack Internet software development (Java, Python, .NET, VUE, JQ) with 3+ years in Rust & Go;10+ years of data solution and architecture experience based on Hadoop, Flink, MySQL, Oracle, SQL Server, MongoDB.
Huge ETH Staking management, digital asset custody, and hundreds of millions in credit fund management experience.