I build tools for alpha research, backtesting, model inference, and developer productivity — with a strong preference for Rust, Polars, Python, and clean system boundaries.
I care about building research infrastructure that is fast, explicit, testable, and close to production.
- ⚡ Performance engineering: Rust-native libraries, low-latency data paths, CPU inference, and async/non-blocking designs.
- 📈 Quant research tooling: factor mining, backtesting engines, trading calendars, data access layers, and analysis workflows.
- 🧠 ML infrastructure: lightweight model runtimes and predictable inference behavior.
- 🛠️ Developer tools: small utilities that remove friction from daily engineering workflows.
|
High-performance non-blocking logging for Rust. Built around a Disruptor-style ring buffer, batch flushing, preallocated records, thread-safe counters, and graceful shutdown semantics.
|
A Rust-powered backtesting plugin for Polars. Designed for strategy evaluation with efficient order matching, long/short position tracking, limit price handling, and a Python API that integrates with Polars DataFrames.
|
|
A high-performance alpha mining framework. Provides an integrated research workflow for factor researchers: trading calendars, data access, factor construction, Rack-based data integration, and Polens factor analysis.
|
A focused Rust inference runtime for official XGBoost Loads supported upstream models, runs CPU prediction, validates model boundaries, and fails explicitly when a model is outside the supported scope.
|
|
A browser-based management UI for Claude Code provider configurations. Replaces manual JSON editing with one-click provider switching, built-in provider presets, API key / OAuth flows, local proxy support, and live synchronization with Claude Code configuration.
|
Core interests: Rust, Python, Polars, PyO3, Maturin, Axum, React, TypeScript, quantitative research systems, backtesting engines, factor analysis, model inference.
| Area | Project | What it demonstrates |
|---|---|---|
| Logging infrastructure | nanolog-rs | Non-blocking Rust logging, ring buffers, batching, graceful shutdown |
| Backtesting engine | polars_bt_extension | Rust + Polars plugin design, order matching, P&L calculation |
| Alpha research platform | alpha-lab | Factor research workflow, data access, trading calendars, analysis tools |
| ML inference runtime | xgboost-rs | Rust-native XGBoost model loading and CPU prediction |
| Developer productivity | cc-switch-ui | Rust backend + React frontend + local config automation |
Make the research loop faster. Make the system boundary clearer. Make correctness visible before performance becomes dangerous.
I like projects that sit at the intersection of research productivity and systems engineering: small enough to understand, fast enough to matter, and explicit enough to trust.