AI-Focused DevOps Engineer

Company: Full Stack

Level: Intermediate (4 years' experience)

Type: Full-Time

Location: Onsite — Cape Town, South Africa

About Full Stack

Full Stack is a software and AI consultancy based in Cape Town. We build products, platforms, and tooling for a broad range of clients — from financial services to retail. Our engineering team runs lean and moves fast, and we are investing heavily in AI-assisted development workflows. We use the tools we build, and we expect everyone on the team to do the same.

The Role

We are looking for an intermediate DevOps Engineer who is genuinely excited about AI tooling — not as a buzzword, but as a daily part of how work gets done.

This is not a pure infrastructure role. You will maintain cloud environments, own deployments, and keep services running, but you will also integrate and support AI-assisted developer tooling, manage LLM inference infrastructure, and help developers stay unblocked. If you have been quietly building confidence with AI coding assistants and want to work somewhere that takes them seriously, this role is for you.

What You Will Do

Cloud Infrastructure

• Provision and maintain Azure resources: App Services, Azure SQL (elastic pools), Key Vault, Entra ID, and resource groups
• Administer Azure DevOps pipelines, variable groups, service connections, and artifact feeds
• Monitor cloud resource spend and flag cost overruns before they happen
• Manage Azure Entra app registrations, service principals, and OAuth 2.0 authentication flows (Device Code, Client Credentials)
• Rotate and manage secrets via Azure Key Vault; enforce the rule that credentials never live in source code

Containerisation & Deployment

• Write and maintain Dockerfile and docker-compose.yml configurations for Python, Node.js, and web application stacks
• Own deployments end-to-end: push, verify, tail logs, confirm health — no silent failures
• Manage nginx as a reverse proxy: SSL termination, upstream routing, and virtual host configuration
• Diagnose container issues: restart loops, failed health checks, port conflicts, network isolation problems

AI Tooling & Developer Platforms

• Deploy and maintain internal AI tooling platforms used by the development team across Windows, macOS, and Linux
• Manage Ollama model deployments (cloud and local), including model availability monitoring and inference endpoint health
• Integrate and maintain third-party AI service APIs: Claude (Anthropic), Google AI / Gemini, ElevenLabs TTS, and others as the stack evolves
• Configure and maintain Claude Code settings, hook definitions, and MCP server registrations
• Troubleshoot AI toolchain failures: model timeouts, token quota exhaustion, broken API connections, authentication errors

Networking & Systems

• Maintain internal network services: DNS, basic VLAN concepts, SSH access management
• Troubleshoot connectivity across Docker networks, on-premises hosts, and cloud endpoints
• Apply firewall rules, port exposure policies, and TLS certificate management (Let's Encrypt and Azure-managed)
• Enforce SSH key-based authentication across all deployment targets

Databases

• Manage Azure SQL databases: schema migrations, connection string management, elastic pool utilisation monitoring
• Perform routine database operations: backups, index maintenance, slow query identification
• Work with ClickHouse or similar analytical databases for data pipeline support
• Understand vector database concepts (pgvector, ChromaDB, or similar) for AI-adjacent workloads

Security

• Enforce secrets hygiene: scan staged files before every commit, audit git history, remediate accidental leaks with git filter-repo
• Maintain .env patterns in .gitignore; ensure every repository has a safe .env.example
• Apply least-privilege principles to all service accounts, Azure RBAC assignments, and API key scopes
• Respond to and document security incidents with clear root-cause analysis and remediation steps

Support & Developer Enablement

• Serve as first-line support for developer tooling issues: broken deployments, failed API integrations, environment misconfigurations, Whisper/TTS pipeline failures
• Maintain cross-platform installation and configuration scripts (Windows PowerShell, bash for macOS and Linux)
• Write and maintain runbooks; if you fix something, document it so the next person (or AI agent) can fix it faster
• Run automated test suites after infrastructure changes; do not ship if tests are broken

What We Are Looking For

Must Have

Cloud & Infrastructure

• Azure (intermediate): comfortable provisioning and managing resources, pipelines, Key Vault, and Entra ID
• Docker (intermediate–confident): Compose stacks, networking, health checks, volume management
• Linux administration: systemd, cron, SSH, bash scripting, process and log management
• Windows administration: PowerShell, Task Scheduler, environment variable management
• nginx: reverse proxy setup, SSL termination, location routing

AI & Developer Tooling

• Active daily use of an AI coding assistant (Claude Code, GitHub Copilot, Cursor, or equivalent) — not just awareness
• Working understanding of LLM concepts: token limits, context windows, model selection trade-offs
• Ability to read and write Python scripts for automation, API integration, and data pipeline tasks
• Comfort integrating with third-party AI APIs (authentication, rate limits, error handling)

Networking

• TCP/IP fundamentals: subnetting, routing, NAT, DNS
• TLS/SSL: certificate types, renewal, and troubleshooting
• Firewall rules: UFW, iptables, or Azure NSG management

Databases

• SQL (intermediate): queries, joins, indexes, basic execution plan reading
• Azure SQL: elastic pools, firewall rules, Azure AD authentication
• Secure connection string management: environment variables, Key Vault references

Security

• Secret scanning and git history auditing
• OAuth 2.0 flows: understand and debug Device Code, Client Credentials, and Authorization Code
• Key rotation and credential lifecycle management

Development Adjacent

• Python (intermediate): write and modify automation scripts independently
• Bash / PowerShell (intermediate): multi-step scripting, API calls, file handling
• YAML: fluent — Compose files, Azure Pipelines, GitHub Actions
• Git (intermediate–advanced): branching, hooks, multi-remote setups, history rewriting when needed
• TypeScript / JavaScript (basic): enough to read configs and not break pipelines

Nice to Have

• Experience with Kubernetes (kubectl, Helm) — we are moving workloads there
• Terraform for infrastructure-as-code on Azure
• Previous work with LLM inference servers: Ollama, vLLM, or cloud model endpoints
• Familiarity with ClickHouse or other column-store analytical databases
• Exposure to ElevenLabs, Runway AI, or other AI media generation APIs
• Understanding of vector search and embedding pipelines (pgvector, FAISS, ChromaDB)
• Familiarity with Tauri or Rust-based desktop applications

The Kind of Person Who Thrives Here

• You reach for an AI tool before writing boilerplate from scratch
• You read logs before guessing — docker compose logs, git log, and cloud monitoring exist; you use them
• You verify before destroying — rollback plans and backups are habits, not afterthoughts
• You write a short runbook after fixing something, because you know it will happen again
• You ask before provisioning cloud resources, because you understand that costs accumulate
• You think about blast radius: who else is affected, what downstream services could break

What a Typical Week Looks Like

Monday:

Check platform health dashboards; review container logs; triage any overnight alerts

Tuesday:

Pipeline work — CI/CD improvements, dependency updates, or a new deployment script

Wednesday:

Developer support — environment issues, broken integrations, tooling questions

Thursday:

Security hygiene — secret scan, key rotation review, access audit

Friday: