I bring seven years of cloud engineering experience to building AI and data solutions around customer workflows. My foundation is GCP, Kubernetes, Terraform, CI/CD, SRE, and MLOps; my focus is turning an operational problem into a working integration with clear success criteria, security controls, and a practical handoff.
Open to Forward Deployed Engineer and Applied AI Engineer roles.
- Discover the workflow: identify the operator, sponsor, baseline, and constraint that shapes the solution.
- Build the integration: connect APIs, exports, and existing systems; choose AI where it helps and deterministic rules where correctness matters.
- Deploy with care: define data boundaries, access controls, evaluation gates, observability, and rollback.
- Measure and hand over: agree on success criteria, write a customer readout, and leave runbooks an operator can use.
These are simulated customer engagements with fictional organizations and sample data. They demonstrate engineering decisions and runnable workflows; they do not claim live customer deployments or measured business results.
| Project | Customer problem and engineering decision |
|---|---|
| Harborline dispatch copilot | Combine shipment exports, tickets, and SOPs into cited dispatch guidance while keeping data local and avoiding external model calls. |
| Payment reconciliation | Reconcile duplicate webhooks and settlement files with idempotent matching; route amount mismatches to an exception queue. |
| Clinic intake | Route after-hours notes with consent checks and minimal audit data; keep the workflow focused on routing rather than diagnosis. |
| Outage desk | Rank storm feeders so a life-safety account is visible without auto-dispatching crews or messaging customers. |
| Field parts | Offline spare lookup that reports stockouts and never creates a purchase order. |
| Vendor review | Flag incomplete packets and bank-detail changes while keeping acceptance with a human reviewer. |
| Customer deployment platform | Self-service Terraform/Helm/Argo files for a fictional customer; prod apply is refused. |
| FDE engagement playbook | Discovery, success metrics, security review, rollout/rollback, weekly readout, and retrospective templates. |
Public labs, not production tenants. Retrieval is local; cluster apply stays off.
| Stage | Evidence |
|---|---|
| ML engineering | Production ML platform · Real-time fraud |
| RAG / eval | PDF RAG · RAG evaluation |
| Agents + MCP | MCP CloudOps · SRE investigator |
| Platform / FDE | Internal developer platform · AI customer deployment |
| Capstone | Enterprise agentic ops |
| Project | What it demonstrates |
|---|---|
| Terraform GCP platform | Multi-environment GCP foundations, Shared VPC, private GKE, IAM, Cloud NAT, observability, and security controls. |
| Customer onboarding platform | Customer onboarding gates that keep a tenant blocked until security checks pass. |
| AI incident investigation | Evidence-backed incident hypotheses without presenting an unconfirmed root cause as fact. |
| upSkilling portfolio | Broader hands-on work across cloud platforms, SRE, MLOps, RAG, automation, and AI infrastructure. |
Tools: Python, FastAPI, GCP, Kubernetes/GKE, Terraform, Docker, Helm, GitHub Actions, Jenkins, GitOps, Ansible, MLflow, Prometheus, and Grafana.
LinkedIn · Resume · FDE playbook
Project guides and local source links are checked on pushes and pull requests. Run locally:
python3 .github/scripts/validate_project_docs.py