PATTERNED AI
Production-ready patterns, architectures, and implementation guides for modern software and AI systems.
FEATURED
LLM Tracing
Trace prompts, model calls, tool calls, latency, and token usage per request.
MCP Server
Expose internal tools through a Model Context Protocol server.
FastAPI Deployment
Wrap model inference behind a typed FastAPI service.
LoRA Fine-Tuning
Adapt a base model with low-rank adapters instead of full fine-tuning.
LLM-as-Judge
Use a rubric-driven model judge to score subjective answer quality.
Basic RAG
Retrieve top-k chunks and ground the answer in context before generation.
Router Agent
Route requests to specialized tools or agents before generation.
LATEST
Prompt Logging
Log prompts and responses with redaction, sampling, and replay metadata.
Cost Tracking
Attribute token spend by tenant, feature, model, and request.
LLM Tracing
Trace prompts, model calls, tool calls, latency, and token usage per request.
Tool Registry
Maintain a versioned catalog of tool schemas, owners, and risk levels.
MCP Client
Connect an assistant runtime to MCP servers with approval and logging.
MCP Server
Expose internal tools through a Model Context Protocol server.
Open Source. Built by AI Engineers.
Contribute patterns, benchmarks, and examples.