• Build Python FastAPI services, data models and reusable connectors for Okta, AWS, ServiceNow and enterprise applications.
• Implement durable execution with Step Functions, SQS and EventBridge. Handle approval waits, rate limits, duplicate events and safe retries.
• Manage PostgreSQL schemas, indexes, query performance and pgvector retrieval. Enforce tenant isolation across APIs, storage and credentials.
• Build model API access for OpenAI, Anthropic and supported AWS paths, including credentials, quotas, timeouts and cost controls.
• Add an analytics pipeline for worker events, traces, tokens, findings and verified outcomes. Evaluate ClickHouse and comparable systems such as BigQuery.
• Create reliable usage records and reconciliation. Prevent retries, duplicate findings and incomplete work from being counted as new outcomes.
• Build deployment pipelines, infrastructure as code, observability and safe release controls across development, staging and production.
• Implement scoped write services, kill switches, audit retention and backup recovery. Test failures and verify that sensitive changes actually took effect.