LAB / 02
Model Router
Route tasks across models using quality, latency, and cost signals.
Independent engineering lab · Online
Building, experimenting,
and thinking about AI.
Independent AI engineering notes and experiments on LLMs, Agents, RAG, model routing and intelligent applications.
LAB://WORKFLOW● LIVE
01understandquery + context
02routemodel + tool
03evaluatequality + signal
system.status
ready_for_experiment10
Labs in the pipeline
03
Engineering notes
∞
Questions to explore
01 / EXPERIMENTS
Selected Labs
Small, inspectable experiments for understanding AI system behavior.
LAB / 01
Intent Router
Turn natural-language requests into intents, entities, and agent routes.
LAB / 03
Prompt Playground
Compare prompt structures and inspect structured responses.
02 / FIELD NOTES
Latest Notes
Architecture decisions, implementation details, and lessons from building.
Do We Still Need Intent Recognition in the LLM Era?
Why intent recognition is becoming a routing contract rather than a single classification step.
From Intent Classification to Agent Routing
A practical architecture for turning user goals into safe, explainable execution paths.
How to Design a Multi-Model Router
Designing model selection around capability, quality, latency, cost, and graceful fallback.
CURRENT FOCUS
Working across the AI stack.
LLMAgentsRAGRoutingMCPEvaluationContext Engineering