These personal projects come from real engineering exploration: breaking down complex problems, establishing clear boundaries, and incorporating execution, auditing, and long-term maintenance into system design.
A local agent system for AI coding, testing, documentation, and research tasks, designed around controllable execution, task auditing, and complex workflow collaboration.
Supports multi-Agent Worker scheduling, task decomposition, and parallel result aggregation.
Incorporates tool permissions, capability authorization, artifact memory, and task auditing into a unified execution model.
Organizes complex task execution chains with Coordinator-Worker collaboration pattern.
A quantitative trading system built with Rust workspace layered architecture, covering backtesting, market replay, simulated trading, and trading operation management.
Covers core modules including strategy, Alpha, portfolio construction, risk control, execution, and OMS.
Provides metrics analysis, account reconciliation, and REST/WebSocket API.
Connects the complete chain from strategy signals to order execution, emphasizing replayability, auditability, and extensibility.
Proxy and management infrastructure for multi-model, multi-account scenarios, unifying account selection, credentials, model availability, and call logs.
Supports unified access to multiple model providers and local models.
Centrally maintains credentials, quota status, model availability, and request logs.
Decouples protocol conversion, account selection, and upstream execution for easy extension to different providers.