Executive prioritisation as a product
Milestone proposals that make the AI opportunity set, expected value and dependencies visible, so sponsors decide together instead of the engineer quietly choosing.
michael@fde:~/cv$ whoami --long
- name
- Michael Myburgh
- role
- Forward Deployed Engineer
- base
- Cape Town, South Africa · on-site or remote
- uptime
- many years shipping production software, still counting
- mode
- embedded with the client, boardroom to codebase
- focus
- AI systems a business can trust
- status
- always up for a chat
- contact
- mp.myburgh@outlook.com
michael@fde:~/cv$ cat profile.md
I'm the engineer you deploy into an organisation when AI needs to move from slideware to production. I work the whole arc in one person: find the problems worth solving, prove them out quickly, then do the unglamorous governance and testing that turns a good prototype into a system people trust with real decisions.
The AI work sits on top of a proper engineering career: many years across healthcare, banking, automotive and logistics, from hands-on IC work to leading large teams. I've modernised legacy platforms, built event-driven systems on AWS and GCP, and carried the budgets and vendor relationships that come with production software.
# how that looks in practice
- Start by listening: interview the people closest to the strategy and the people closest to the work, then trace every request back to the problem underneath it
- Prototype in days, sized to the uncertainty. A small experiment should answer the expensive question first
- Promote only validated ideas to production grade, with the diligence that word implies
- Keep progress visible to the people funding the work. Visibility is part of the job
michael@fde:~/cv$ ls capabilities/
discovery/stakeholder interviews, workflow & friction mapping, opportunity framing
prototyping/small experiments on real data, built in days and sized to what they need to prove
production-ai/governed connectors, eval & regression harnesses, query-reliability engineering
data-governance/approved views over raw access; partnering with the teams that own truth
security/least-privilege by design; an incident must never originate from my work
adoption/demos that land, and taking the human side of AI change seriously
michael@fde:~/cv$ tail -n 2 delivery.log
ONGOING
ALWAYS
Own infrastructure, end to end
DNS, TLS, VPN tunnels, firewalls, monitoring. This site runs on a server I administer myself.
michael@fde:~/cv$ stack --list
languagesJava & Spring Boot (the deep end), Python, Go, TypeScript, SQL (PostgreSQL, Oracle, MySQL)
ai/llmagentic systems, tool use & MCP connectors, eval & regression harnesses
cloudAWS (Lambda, SageMaker, EKS, Step Functions), GCP (Cloud Run, Pub/Sub, BigQuery), Kubernetes, Docker, Terraform, AWS CDK
architectureevent-driven microservices, multi-tenant platforms, serverless (Quarkus/GraalVM), legacy modernisation
frontendAngular, React, and enough design sense to build this page
qualityJUnit, Spock, Cypress, TestContainers, SonarQube, CI/CD with Jenkins & GitHub Actions
securityOAuth2 / OIDC, OWASP, POPIA & GDPR, least privilege as a habit
opsLinux, networking & VPNs (WireGuard), nftables, TLS, Grafana & Kibana
humanstakeholder management, discovery interviews, demos, leading large teams, vendor & budget management
michael@fde:~/cv$ cat principles.txt
- Discovery before solutions. Trace every request to the problem behind it.
- Match engineering maturity to certainty: prototype fast, productionise deliberately.
- A proof of concept that never ships can still pay for itself.
- Visibility is part of delivery.
- Never give AI unrestricted access to business-critical systems or data.
- Build on governed sources of truth; partner with the teams that own them.
- Make prioritisation explicit and shared instead of quietly absorbing every request.
- Adoption is human: stay approachable, and take people's concerns seriously.
michael@fde:~/cv$ contact --open
locationCape Town, ZA · UTC+2 · remote-friendly
statusalways open to a chat
responsefast. embedded engineers answer their mail.
# this prompt is real. try: help
michael@fde:~/cv$