Regulated EU Fintech, Insurance & HealthtechBook a call
Case studies
Work that removed real friction.
A few engagements where the win wasn't a new feature — it was deleting the operational drag everyone had quietly accepted.
Much of my work sits under NDA, so some details here are described in general terms and some projects aren't shown at all. Happy to talk through specifics on a call.
Compliance
SOC 2 certified in 6 months — 3 months early
SOC 2 certification in 6 months at Lynx.MD
Situation
A healthcare data company needed SOC 2 certification within 9 months to close enterprise deals — with no compliance program in place, no owner for the process and controls that existed only informally inside engineering.
Action
I owned the process end-to-end: gathered the requirements, coordinated across security, legal and engineering, and implemented the technical controls myself — encryption at rest and in transit, and access-role separation across AWS RDS and the surrounding services — so the evidence came from how the system actually ran, not from a spreadsheet.
Result
6 months
From zero program to certified
3 months
Ahead of the deadline
Enterprise
Sales unblocked
Certified three months ahead of plan
Unblocked enterprise deals waiting on SOC 2
Encryption and role separation enforced across AWS
Security program embedded in day-to-day engineering
Compliance sticks when the controls are part of how engineers already work — not a parallel process bolted on before the audit.
Automation
Environment setup cut from ~2 weeks to ~1 hour
From two weeks of manual setup to one automated hour
Situation
In one of my previous roles, every new engineer was required to provision a full customer environment as part of onboarding. The intention was good: build deep, end-to-end understanding of the system. In practice, it meant navigating ~65 manual configuration steps across database, AWS accounts, Google services, permissions and tightly coupled environment settings. A single setup typically took close to two weeks — and the result was often inconsistent, with environments breaking on subtle misconfigurations or missed dependencies.
Action
Rather than treating this as a "learning exercise", I saw it as a systems problem. The onboarding process wasn't scaling: it introduced unnecessary operational load, inconsistency across engineers and a significant amount of avoidable friction. I built an end-to-end automation layer using Terraform and Python to replace the manual workflow — orchestrating infrastructure provisioning, cross-system configuration, dependency ordering and validation checks to guarantee consistent, repeatable environments.
Result
~2 weeks → ~1 hour
Environment provisioning time
~65 → 0
Manual configuration steps
100%
Repeatable, validated environments
Eliminated manual configuration errors
Standardized environment provisioning across engineers
Reduced operational overhead and troubleshooting time
Improved onboarding consistency and time-to-productivity
The highest-leverage improvements often come from removing friction in systems people have normalized — not just from building new features.
Observability
One logging pipeline for thousands of engineers' workloads
Kubernetes operators for centralized logging at a global payment solution
Situation
Log data from Kubernetes pods was fragmented across services, making it slow to trace issues at a global payment solution's engineering-org scale.
Action
I deployed Kubernetes operators alongside FileBeat as part of the Artifactory/logging migration, standardizing log collection and shipping across pods.
Result
Centralized
One logging pipeline across pods
Faster
Root-cause tracing
1000s
Engineers' workloads covered
Standardized log collection across Kubernetes pods
Removed fragmented, per-service log handling
Cut time to trace issues across services
Delivered as part of the wider Artifactory migration
Observability pays off when collection is a platform default — not something every team wires up on its own.
Platform Engineering
You don't notice it until it breaks.
Artifact management migration to JFrog Artifactory at a large payments company
Situation
Every build. Every deployment. Every dependency. All of it flows through your artifact management layer. And most people only notice it exists when it breaks. At a large payments company, a fragmented setup of legacy Harbor and Nexus instances sat underneath everything engineers shipped.
Action
We replaced that fragmented setup with JFrog Artifactory, and I built the logging and monitoring layer that tracked artifact resolution, deployment flows and failure modes across on-prem VMs and Kubernetes. When you're migrating something this central, you can't afford silent failures — you need to know exactly what broke, where, and why, before anyone opens a ticket.
Result
99.6%
Artifact resolution success rate
0.27%
Error rate under real-time visibility
Harbor + Nexus → 1
Consolidated artifact layer
Turned a high-risk migration into a controlled one
Unified Docker, Maven, npm, PyPI and Helm repositories
Logging, monitoring and alerting across on-prem VMs and Kubernetes clusters
Failure modes visible before tickets were opened
Modernizing infrastructure inside an established organization isn't really a tooling problem. It's a visibility problem — you can't safely change what you can't see. Sixteen years in, I still find that's where most of the value hides.
