design AI agent architecture including planning boundaries, execution loops, memory and state strategy, tool routing, approval gates, retries, delegation, and halt behavior.
design observability for AI agents and workflows including traces, prompts, model calls, tool calls, retrieval events, approvals, errors, eval probes, cost, latency, and safety signals.
orchestrate AI engineering workflows from capability intent through model, prompt, tool, agent, retrieval, eval, safety, inference, observability, release, and incident stages using connector-grounded evidence, workflow packets, stage advancement, and halt…
triage AI production incidents involving hallucination spikes, safety failures, prompt injection, tool misuse, data leakage, model regressions, cost spikes, latency degradation, eval regressions, or user harm reports.
assess readiness to release AI capabilities across requirements, evals, safety review, red-team status, inference ops, observability, rollback, docs, support handoff, and owner approval.
review AI capability risks including misuse, policy compliance, privacy, security, hallucination harm, data leakage, autonomy, tool-use risk, user impact, and mitigations.
optimize AI system cost and latency using model routing, caching, prompt compression, context pruning, batching, streaming, parallelism, retrieval tuning, and fallback tiers while preserving quality and safety gates.
plan and review AI datasets for source selection, labeling, balancing, privacy, deduplication, train and eval splits, drift, provenance, consent, and retention.