About Epistates
We're an AI engineering company that builds production-grade solutions for businesses. Our team brings over a decade of experience in distributed systems and fintech to every engagement.
Engineering AI That Actually Ships
Epistates was founded by engineers who spent years building high-frequency trading systems, distributed infrastructure, and real-time data platforms. We brought that same rigor — performance-obsessed, security-first, production-grade — to AI.
We don't just integrate APIs. We build the infrastructure that powers AI: MCP protocol SDKs, LLM fine-tuning frameworks, multi-agent orchestration systems, and the developer tools that tie them together. Our open source projects are used by thousands of developers worldwide.
When a business comes to us, they get the same engineering team that built TurboMCP, PMetal, and TurboVault — working directly on their problem with the same quality bar we hold for our own products.
What We've Built
Our flagship open source projects power AI infrastructure used by developers and enterprises worldwide.
TurboMCP
v3.0.6Production-ready Rust SDK for the Model Context Protocol. 25-crate modular architecture with zero-boilerplate macros, 5 transport protocols, and OAuth 2.1 security.
View ProjectPMetal
v0.3.13High-performance LLM fine-tuning framework optimized for Apple Silicon. LoRA/QLoRA/DoRA, knowledge distillation, GRPO reasoning training, and 20+ model architectures. We support training on any hardware.
View ProjectTurboVault
v1.2.9MCP server that transforms Obsidian vaults into intelligent knowledge systems. 44+ tools, BM25 search, link graph analysis, and sub-100ms performance.
View Projecttreemd
v0.5.9Markdown navigator with tree-based structural navigation, jq-like query language, syntax highlighting for 50+ languages, and vim-style keybindings.
View ProjectWhere We Come From
The same engineering discipline that powers Wall Street trading systems — applied to AI.
Fintech & Trading
- High-frequency trading infrastructure
- Real-time market data pipelines
- Risk management systems
- Algorithmic execution engines
Distributed Systems
- Low-latency microservice architectures
- Event-driven processing at scale
- Multi-datacenter deployment
- Zero-downtime production operations
AI & Machine Learning
- LLM fine-tuning & deployment
- MCP protocol infrastructure
- Multi-agent orchestration
- On-premises ML training pipelines
How We Work
Ship Production Code
We don't build demos. Every solution we deliver runs in production with real users, real data, and real uptime requirements.
Own Your AI
Your models, your data, your infrastructure. We build solutions you control — no vendor lock-in, no per-token fees, no third-party access.
Open Source First
Our core tools are open source and battle-tested. You benefit from community-hardened infrastructure with enterprise support.
Move Fast, Ship Right
Weeks, not months. We move quickly without cutting corners — rigorous testing, comprehensive docs, enterprise security from day one.
Corporate Information
Let's Build Something Together
Whether you need custom AI agents, MCP infrastructure, or LLM fine-tuning — we bring a decade of engineering discipline to your problem.