AI engineering
AI systems that hold up outside the demo.
We build AI systems the same way we audit software: a fixed written quote, senior-engineer hours stated up front, and an independent review before anything goes live.
Search and ranking
Relevance models for product search, autocomplete and recommendations; test sets and graded judgements to measure whether ranking improved.
Agents and tool servers
Chat assistants grounded in your own data and tools; MCP servers so ChatGPT and Claude can use your systems safely, with rate limits and scoped access.
Evaluation and LLM judges
Benchmarks built from your real tasks; small judge models that grade outputs consistently and cheaply.
Model development
Fine-tuning small open models with LoRA for a narrow task, trained on our own GPUs or yours; we ship the evaluation alongside the model.
Self-hosted deployment
Open models served on hardware you control, for data that must not leave your network.
Human accountability
Honest about AI
Our work is heavily AI-assisted, and we say so. Every quote states the senior engineer hours we commit to your code, a senior engineer reviews every finding, and our warranty is measured in the same hours.
Our AI-assisted processStart a conversation