AI Platform Engineer · DevOps & MLOps · Istanbul, Turkey
I build the paved roads between ambitious AI ideas and production-ready systems.
╭─ system.identity ─────────────────────────────────────────────╮
│ name Rojin Ebrahimi │
│ focus LLM gateways · agentic workflows · cloud native │
│ principle secure + observable + repeatable by default │
╰───────────────────────────────────────────────────────────────╯
I work where AI infrastructure meets developer experience: creating gateways, orchestration, and platform foundations that help teams use LLMs safely and productively at scale. My DevOps background means I care as much about reliability, latency, and operability as I do about what a model can do.
NOW |
IMPACT |
OPEN TO |
|---|---|---|
| AI Platform Engineer @ Snapp | Secure agentic delivery paths for 1,000+ developers | AI Platform Engineering · MLOps · DevOps |
| ▣ AI Gateway Architecture Designing internal LLM access that is reliable, governed, and secure. |
◇ Agentic Workflows Integrating custom AI agents and MCP servers for safe, useful tool calling. |
◉ Developer Enablement Standardizing the paths that help teams ship AI without operational debt. |
AI Platform & MLOps
MLflow Evidently AI LLM Gateways MCP Agentic Tooling AI Orchestration Custom AI Agents
Cloud-Native & Delivery
Kubernetes RKE2 Docker Helm Terraform Ansible Argo CD GitLab CI/CD GitHub Actions Python Go Linux Shell
Reliability, Security & Data
Prometheus Grafana Zabbix ELK Stack Sentry Kafka RabbitMQ Redis ClickHouse Vault Consul
| Repository | Signal |
|---|---|
| ansible-playbooks | Reusable automation for installation and configuration work. |
| python-automation | Practical utilities that remove repeatable work from everyday workflows. |
| portfolio | A Go portfolio that documents my technical direction in public. |
The projects I’m most excited to build in the open:
flowchart LR
A[MCP Policy Lab] --> B[LLM Platform Scorecard]
B --> C[GitOps for AI Services]
C --> D[Agent Failure Mode Playground]
- MCP Policy Lab — a reference platform for permissions, audit trails, and safe tool-calling boundaries.
- LLM Platform Scorecard — an opinionated dashboard for gateway health, cost, latency, quality, and model usage.
- GitOps for AI Services — production-minded templates for models, agents, evaluations, and observability.
- Agent Failure Mode Playground — reproducible failure scenarios with practical reliability patterns.
I’m the founder and host of Enguru, a bite-sized English-learning podcast for clearer, more confident communication. Away from a terminal, I figure skate, practise yoga, paint, and play guitar. I’m also improving my German—currently B2.