Software engineer building agentic AI systems and the infrastructure that makes them measurable — cost attribution, evaluation, observability and guardrails.
Cost attribution and spend observability across multiple LLM providers, broken down per feature and per tenant, with budget alerting and anomaly detection.
A routing layer that picks the model tier per request from task complexity, latency budget and cost ceiling.
Built the product's own AI layer with an eval harness gating every release and quality metrics tracked per commit.