Turning cloud reliability, cost discipline, and observability into a competitive advantage — at the intersection of platform engineering and strategic leadership.
Four stories that define how I approach platform engineering — with urgency, precision, and long-term thinking.
Our AWS spend was running $80–110K over budget every month — peaking at $782K against a $675K target. Rather than treating this as a finance problem, I treated it as an engineering one.
I led a systematic cost reduction program: rightsizing compute instances to match actual workload profiles, migrating eligible jobs to Spot Fleet, replacing on-demand spend with Reserved Instances and Savings Plans, and aggressively eliminating idle and orphaned resources across accounts.
The outcome: total spend fell from a $782K peak to $665K — the first month we came in under our $675K budget. Savings are now structural, backed by real-time guardrails that keep every team accountable to their cloud footprint.
The engineering org runs ~2,000 Jenkins jobs per day across a complex build, test, and deploy cycle. Debugging failures meant context-switching between multiple platforms. Worse, CI pipelines ran overnight US time — meaning the EU team waited nearly a full day for feedback on every commit.
I implemented Datadog CI Pipeline Visibility to create a unified view across all Jenkins jobs — build, test, and deploy — replacing manual monthly spreadsheets with daily SLOs on execution time, queue time, and blocked time.
The result: ~50% reduction in EKS CI costs through workload optimization, and CI feedback collapsed from ~1 day to near real-time for both US and EU teams — without duplicating infrastructure. Datadog featured this work in their official case study.
Cost visibility was scattered across the org: AWS Cost Explorer for raw billing data, nOps for cost reporting, and separate systems for performance and infrastructure monitoring. Teams were constantly context-switching, and cost data lived in a silo disconnected from the engineering decisions that drove it.
I led the strategic migration of cost observability into Datadog — consolidating cost reports, CI metrics, infrastructure performance, and alerting into a single organizational data ecosystem. By putting cost data next to the performance metrics that cause it, teams could no longer ignore the relationship between the two.
Today, engineering teams across the org operate from real-time cost and performance dashboards in a single platform — making cost a first-class engineering concern, not a quarterly finance review.
Incidents were being resolved but not learned from. Root-cause analysis was inconsistent, action items fell through the cracks, and the same failure modes kept resurfacing across quarters — draining engineering time and eroding SLO budgets.
I built a structured incident and postmortem framework — standardizing severity classification, blameless RCA processes, and systematic tracking of corrective actions through Datadog, Snowflake, and Tableau dashboards. Every postmortem became a closed loop with measurable follow-through.
The result: 35% reduction in repeat incidents and a 25% improvement in mean time to recover (MTTR) — turning postmortems from a compliance exercise into a genuine reliability engine.
External validation that the thinking holds up beyond the org chart.
The beliefs that guide every architectural decision, team conversation, and tooling choice.
Not just used — architected, operated, and improved at scale. Every tool here represents real production ownership across cloud, observability, and platform engineering.
Automation, runbooks, and tools built from real platform problems — things I've packaged up for others to use or adapt.
A collection of Python scripts, Shell automation, and documented runbooks for identifying rightsizing candidates, automating Savings Plan analysis, and tagging unused resources — built from the real $100K/mo savings initiative.
A ready-to-import Datadog dashboard template for tracking CI pipeline health at scale — execution time, queue time, blocked builds, and P95 latency across Jenkins and GitHub Actions. Inspired by the Datadog case study.
A practical starter kit for engineering teams new to SLOs — includes a template library, error budget calculator, and a guide on setting initial availability and latency targets without overcomplicating the rollout.
Currently packaging up Terraform modules for multi-cloud cost tagging and a FinOps maturity assessment template. Watch this space — or connect on LinkedIn.
Whether you're thinking about cloud cost strategy, platform team structure, or what it looks like to build observability culture at scale — I'm always open to a conversation.
Connect on LinkedIn