Architect and Senior SRE focused on observability, distributed systems, and developer tooling. I write Python, Bash, and IaC and I care about keeping things running.
Fragmented logging and tracing across multiple vendor tools made cross-service debugging slow and expensive.
Outcome › In-progress — unifying logs, metrics, and traces under a single Grafana experience for personal infrastructure.
Manual My Oracle Support (MOS) Service Request triage and stakeholder notifications slowed mean time to resolution for critical Sovereign Cloud incidents.
Outcome › Automated SR creation, severity triage, and notifications via Python + OCI APIs; measurably reduced MTTR across low-side and high-side government realms.
Manual Change Advisory Board review created a deployment backlog for OCI service teams, slowing safe release velocity.
Outcome › Engineered automation pipelines that eliminated manual change request processing, cut the review queue, and accelerated deployments for the Sovereign Cloud org.
Production outages required manual triage that extended downtime windows and risked SLA breach across distributed services.
Outcome › Cut downtime 40% with automated monitoring, runbooks, and recovery tooling; sustained 99.99% availability and reduced manual operational effort by 50%.
On-call isn't just alert response — it's building a blameless, high-trust team that learns from production incidents fast.
P99 latency measures the slowest 1% of your requests — a better health signal than averages for catching tail latency issues in production.
Which cloud dollars are doing real work, and which ones are just vibing? An SRE's guide to spending less without breaking everything.