Plan workspace architectures; implement SSO/SCIM integration, role-based access controls (RBAC), and policy guardrails.
Establish usage analytics, define custom GPT governance (actions, connectors, approval workflows), and create runbooks and SOPs.
Design scalable patterns for Assistants with multi-tool orchestration, structured JSON outputs, function/tool calling, Files/Batch for bulk operations, and Moderations and Embeddings for retrieval workflows.
Implement OpenAI-centric RAG pipelines (chunking, embeddings, indexing) with groundedness checks, prompt test suites, red-teaming, and defined cost/performance SLOs.
Own end-to-end fine-tuning processes — dataset curation, training and evaluation, bias and quality checks, rollback/versioning, and telemetry for tuned models.
Implement OpenAI Observability and OpenTelemetry for performance monitoring; add token/cost telemetry, retries/backoff, idem...