AI Engineer

Location: South Florida (Miami-Dade / Broward / Palm Beach) — Local candidates only

Type: Full-Time or Consultant

Compensation: Salary + Benefits

About the Role

We're looking for a hands-on AI Engineer to take over, harden, and scale our autonomous agent infrastructure. Our v1 is designed, built, and running — this role is about reconstituting and refactoring that system into an architecture built for growth. You'll inherit a working stack, learn it inside and out, then re-architect where it makes sense: cleaning up technical debt, improving reliability and observability, and planning capacity for significantly higher throughput.

This is not a prompt-engineering or API-wrapper role, and it's not greenfield either. It requires someone who can read and reason about an existing system before rebuilding it — comfortable across the full stack, from hardware and inference servers to agent harnesses and production business workflows. You'll work directly with leadership and have significant autonomy over technical decisions.

What You'll Do

Own the existing agent infrastructure end-to-end — audit the current v1 architecture, document it, and drive the refactoring roadmap toward a scalable v2
Refactor and extend agent harnesses — orchestration logic, tool integrations, memory/state management, scheduling (heartbeats/cron), and guardrails — improving what exists and rebuilding components where the design doesn't scale
Operate and extend our deployments of open-source agent frameworks including OpenClaw, Paperclip, and Hermes Agent — custom skills, plugins, multi-agent orchestration, and version upgrades without breaking production
Optimize LLM prioritization and routing — matching tasks to the right model based on cost, latency, capability, and context requirements across multiple providers, and improving on the current routing logic
Maintain and scale offline/local LLM deployments (Ollama, vLLM, llama.cpp, or similar) on self-managed hardware, including quantization, context tuning, and throughput optimization
Plan hardware capacity for growth — GPU selection, server builds, cooling, networking, and monitoring as inference load increases
Strengthen integrations between agent systems and production infrastructure: databases, APIs, messaging platforms, and internal tooling
Build out observability, cost controls, budget enforcement, and audit logging so the system stays reliable and accountable as it scales
Evaluate new models, frameworks, and tooling as the ecosystem evolves — and know when to refactor, replace, or leave well enough alone

Required Qualifications

Located in South Florida and able to work on-site / hybrid — no relocation, no fully remote candidates
3+ years of software engineering experience (Python and/or Node.js/TypeScript strongly preferred)
Demonstrated hardware experience — building/maintaining servers or GPU workstations, Linux system administration, networking fundamentals
Hands-on experience with agent harnesses and frameworks — you've deployed and customized tools like OpenClaw, Paperclip, or Hermes Agent, and you understand harness internals well enough to have built agent loops from scratch (tool calling, state persistence, sub-agent spawning, error recovery). We need that depth to refactor an existing system, not to start a new one
Experience inheriting and refactoring production systems — you can read someone else's architecture, identify what to keep vs. rebuild, and make changes without downtime
Experience with LLM prioritization/routing — multi-model architectures, provider failover, cost/latency tradeoffs
Experience configuring offline/local LLMs — running open-weight models (Llama, Qwen, DeepSeek, Mistral, Hermes, etc.) on local hardware, including quantization formats (GGUF, AWQ), inference servers, and VRAM/context management
Strong Linux/CLI fluency — shell scripting, systemd/LaunchAgents, Docker, environment and secrets management
Comfort working independently in a fast-moving environment with minimal oversight

Nice to Have

Experience with MCP (Model Context Protocol) servers and tool integrations
Fine-tuning or LoRA training experience on open-weight models
Experience with vector databases and RAG pipelines
PostgreSQL and general database administration experience
Background in ad tech, performance marketing, or high-volume data pipelines
Experience with messaging-platform integrations (Telegram, Slack, Discord, WhatsApp)
Contributions to open-source AI/agent projects

What We Offer

Competitive Compensation + Health/Dental/Vision Benefits
Unlimited PTO
Real ownership over architecture and tooling decisions
Budget for hardware and experimentation
Small team, no bureaucracy — your work ships and matters immediately

How to Apply

Send your resume along with links to relevant work — GitHub repos, agent projects, homelab setups, or a short write-up of an agent system you've built or refactored. Bonus points if you can describe a system you inherited and how you improved it. We care more about what you've shipped than where you've worked.

michael@magicmediagroup.co

Magic Media Group LLC AI Jobs

Latest AI Jobs at Magic Media Group LLC

Artificial Intelligence Engineer