Rust Performance Engineer | AI Tooling Architect
I build high-performance systems that solve real problems. My work spans from SIMD-accelerated parsers to AI agent infrastructure—always with a focus on measurable impact.
Core contributor to foundational AI infrastructure:
| Project | Contributions | |
|---|---|---|
| modelcontextprotocol/rust-sdk | Elicitation support, context-aware completion | |
| restatedev/restate | Shell completion for restatectl | |
| zed-industries/extensions | Deps extension |
Note
Tools that make AI agents cheaper and smarter.
| Project | What it does | Impact | |
|---|---|---|---|
| mcp-execution | MCP servers → standalone TypeScript tools | 98% token savings | |
| claude-plugins | 8 specialized Rust agents for Claude Code | Full dev lifecycle coverage | |
| mcpls | MCP ↔ LSP bridge | Semantic code intelligence for AI | |
| pjs | Priority JSON Streaming + SIMD | 6x faster than standard parsers | |
| tap-mcp-bridge | Visa TAP + MCP integration | Secure payment auth for AI agents |
Tip
Start with mcp-execution it's the foundation for token-efficient AI workflows.
Note
When standard libraries aren't fast enough.
| Project | What it does | Impact | |
|---|---|---|---|
| exarch | Memory-safe archive extraction, TAR/ZIP | Zero CVEs · replaces vulnerable libs | |
| fast-yaml | YAML 1.2.2, zero unsafe, built-in linter | 14x faster than PyYAML | |
| feedparser-rs | RSS/Atom/JSON Feed + all extensions | 213 MB/s · 94x faster | |
| pyhdb-rs | SAP HANA driver with Arrow + async | Polars/pandas integration | |
| scrape-rs | HTML parsing + CSS selectors, SIMD | 50x faster than BeautifulSoup |
Important
All have Python bindings — drop-in replacements for existing code. exarch, fast-yaml, feedparser-rs, and scrape-rs also have Node.js bindings.
Note
Tools I wish existed.
| Project | What it does | Impact | |
|---|---|---|---|
| deps-lsp | Universal dependency management LSP | Cargo + npm + PyPI | |
| deps-zed | Zed extension for deps-lsp | Version hints + diagnostics | |
| helix-trainer | FSRS spaced repetition for Helix | 30% fewer reviews than Anki |
Note
Completed projects no longer in active development.
| Project | What it was | |
|---|---|---|
| vkteams-bot | VK Teams Bot ecosystem: CLI + MCP + vector storage | |
| vkteams-bot-clj | VK Teams Bot implementation |
Building infrastructure for AI agents that doesn't waste context on ceremony.