Bob Ulrich

Remote

• › Shipped a production multi-agent system that runs thousands of deployment investigations daily, 24/7: agents on a durable workflow engine route each task to the right Claude model, verify failures against live evidence, and escalate low-confidence cases.
• › Cut LLM cost per investigation 44-57% (~62K to ~30K tokens) with a deterministic gate that skips a redundant model call on ~70% of investigations, 10-15x tool-output compression, and per-run cost attribution built from scratch.
• › Own reliability for an internal developer platform deploying 300+ services across commercial and GovCloud (FedRAMP) environments, holding a 99%+ deployment success rate on AWS (EKS, Lambda, Step Functions).
• › Raised SLI accuracy from a 42% noise rate to actionable signal by classifying pipeline failures as platform vs service-team issues and calibrating SLO alert thresholds against 60-day statistical baselines.
• › Designed the evaluation methodology for AI code review at enterprise scale, calibrating security, compliance, performance, and maintainability checks against thousands of production pull requests before rollout to developers.

Phoenix, AZ

• › Promoted to Principal SRE on Substrate Compute, operating 10,000+ hypervisors with capacity planning for the local storage fleet.
• › Eliminated 90% of manual deployment toil and cut rollback incidents 75% by replacing tmux/script workflows with a Terraform infrastructure-as-code pipeline.

Seattle, WA

• › Cut on-call escalations 30% across a 15,000+ hypervisor fleet through trend analysis, runbook automation, and targeted toil elimination.
• › Operated QEMU/KVM compute with iSCSI storage at 99.9%+ availability.

Redmond, WA

• › Led a Kubernetes migration that cut deployment time 50% and infrastructure costs 30%.
• › Built a CI/CD pipeline that took release cycles from weeks to hours and raised deployment frequency 10x.
• › Reduced mean time to detection 60% by implementing an EFK (Elasticsearch, Fluentd, Kibana) logging stack.

Redmond, WA

• › Operated consumer cloud storage serving hundreds of millions of users and contributed to the Azure microservice migration that reduced deployment complexity.

Redmond, WA

• › Maintained a 99.99% SLA for 100M+ monthly active users on the 24x7 SRE team.
• › Cut catastrophic storage failures 67% in one year through process improvements.
• › Reduced farm-wide update time from days to hours (3x faster) by re-engineering the SharePoint update process.

Redmond, WA

• › Migrated 4,000+ servers to hardware SSL offloading, cutting TLS handshake time 50%.