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.
