Operations¶
The operator surface: deploying Bernstein, running it day to day, and keeping it healthy, cheap, secure, and auditable.
-
Deploy and scale
Docker Compose, Helm, cluster mode, fleet, air-gap installs.
-
Run and supervise
Run, schedule, resume, replay, fork, and review agent work.
-
Reliability
Troubleshooting, retries, auto-heal, stall escalation, disaster recovery.
-
Cost and performance
Budgets, cost-aware scheduling, performance tuning.
-
Observability
Instrumentation, deterministic replay, telemetry, trends.
-
Merge and review automation
Merge queue, autofix daemon, review responder, coverage ratchet.
-
Evaluation and calibration
Incident-to-eval synthesis, A/B runner, calibration.
-
Security and identity
Credential scoping, secrets, hardening, capability matrix.
-
Compliance and audit
EU AI Act, SOC 2, audit log, lineage export.