At Orcrist, led a GitOps platform for law enforcement on GKE. At E.ON, led AI into production for 8M+ energy customers with 2M+ monthly LLM interactions. At Allianz, cut cloud provisioning from over a year to days.
Project constraints, clear deliverables, and the steps we took to get there.
Ask me anything. Answers drawn from the work on this site.
Platform migrations, team scaling, enterprise architecture. Answers drawn from the work on this site.
Start a conversation →Based on project data.Three models, each shaped around your team and your stage.
A day with your leadership and engineers to map where AI delivers value and aligns with your regulator.
An outside perspective to frame the problem, pressure-test assumptions, and provide a clear execution plan.
I own the roadmap, discovery, and operating rhythm while your engineers focus on delivery.
Every engagement follows the same arc: listen first, ship to production, then hand over a system your engineers understand inside out.
I embed in your Slack, standups, and incident calls and spend the first week listening — mapping the dependency graph and how decisions get made — then frame the core problem. We surface security and compliance constraints (GDPR, BSI, KRITIS) immediately so they guide the architecture rather than blocking the launch.
I own the roadmap end to end and we ship to production — evals, observability, and safe rollout included. A weekly operating cadence and KPI dashboards keep progress visible to everyone.
We prove the outcome against the KPIs we set. I leave behind a documented operating rhythm and playbooks, ensuring your engineers have full control of the platform.
What colleagues and clients say after the engagement ends.
“As project lead for the observability program, I valued that Tungi could align platform teams and business stakeholders around one playbook. He structured the rollout of OpenTelemetry and New Relic so that each new domain onboarded faster and with fewer surprises. Incident reviews are now grounded in shared journeys and SLOs, not just raw logs, which noticeably reduced detection and resolution times.”
Honest answers to the things cautious buyers want to know.
A 30-minute call to see if there's a fit. No pitch deck.
