Résumé

The work, on one page.

Agentic Engineer building multi-agent LLM systems on Claude, MCP servers, and production AI infrastructure that runs day in and day out, under real load. Targeting Forward-Deployed Engineer roles.

Coral Springs, FLrbonnaig@dreambigai.iolinkedin.com/in/roody-bonnaiggithub.com/AlphaTronOmegaRoo

Summary

I build AI systems that run in production, day in and day out, under real load. For the past year I've built and operated a multi-agent system on Claude: around seventeen LLM-driven agents running continuously across two cloud servers and a local workstation. It runs on Claude today and is being built to be model-agnostic, tied together by a bespoke memory layer of my own: version-controlled files plus a Qdrant vector database for retrieval (RAG), reached through a gateway I can repoint, so nothing is tied to one vendor. When something breaks at two in the morning, I'm the one who finds out why and fixes it so it doesn't break the same way twice. That's the part I love most. The work is forward-deployed in shape: I ship MCP servers and agent skills, and the deployment patterns that make the thing reliable enough to hand off to someone else. More than thirty years in customer service came before the code: two decades-plus as a server in food and beverage, retail customer service before that, and most recently hospitality at The Boca Raton Resort. So the non-code half of the job is second nature. I can sit down with someone who doesn't code, get to what they actually need, and build it. Right now I'm an IT Support Analyst at The Boca Raton Resort, finishing an Applied AI degree at Miami Dade College, and working through the Anthropic Claude certification track.

What I Bring

  • A multi-agent LLM system on Claude, running continuously for about a year across two cloud servers and a local workstation. The closest thing I have to forward-deployed work, and it's mine end to end.
  • 20+ MCP servers integrated, several written from scratch, plus 64+ custom Claude Code skills. The exact artifacts the role ships.
  • A full-stack AI SaaS (SupplyScout) built with Claude, AWS Textract, and PostgreSQL, with an evaluation and confidence-scoring loop and a human-review step.
  • Embedded enterprise supportas an IT Support Analyst inside a Forbes Five-Star resort's live production systems: white-glove, on-the-floor, under real stakes.
  • Repeatable deployment patterns codified as a library of production safety gates and named engineering rules, reused across every project.
  • Thirty-plus years in customer service, building long-term relationships front-of-house and on the floor.
  • Translation between non-technical people and technical systems. The recent skill I picked up as an IT analyst, turning what people actually need into the systems they depend on.

AI & Agentic Engineering

Multi-Agent AI System | Aug 2025 – Present

A production environment of coordinated LLM agents (internally, "the Colony").

  • Designed and operate a system of ~17 specialized AI agents on Claude, running across two cloud servers and a local workstation, with parallel task execution, autonomous operation under human oversight, and shared memory plus retrieval over a Qdrant vector database (RAG).
  • Built and integrated 20+ MCP (Model Context Protocol) servers that let the agents act on databases, APIs, file systems, and the browser. Servers I wrote from scratch include audio analysis (audio-ears), image/video generation (ComfyUI), GPU monitoring, YouTube automation, and a local-inference delegation layer for Ollama; integrated servers cover PostgreSQL, Qdrant, Playwright, n8n, Obsidian, and Perplexity.
  • Shipped 64+ custom Claude Code skills across automation, content production, DevOps, and system administration.
  • Wrote a library of production safety gates (operator-default checks, verification gates, destructive-action guards, stash auditing, backup-integrity verification). Reusable across every project.
  • Built my own memory layer, independent of any model vendor: version-controlled file contracts (durable STATE / MEMORY / log) plus Qdrant vector RAG, reached through a model gateway I can repoint. A self-improvement loop lets agents capture lessons each session and rewrite their own instructions.
  • Result: roughly 60+ hours a month of manual work automated.

SupplyScout: AI Restaurant Procurement Platform | Sep 2025 – Present

Full-stack AI SaaS, concept to a production-ready v2.0 (built and tested; market-researched, not yet live with paying customers).

  • Built with Next.js 14, TypeScript, PostgreSQL (Supabase), Stripe, and AWS Textract.
  • Architected a 35-table PostgreSQL database with row-level security; migrated the data layer from MongoDB.
  • Built an invoice-reading pipeline on AWS Textract with an evaluation and confidence-scoring loop and a human-review step, at 95%+ field-extraction accuracy.
  • Integrated Stripe 3-tier billing ($79 / $199 / $399 per month) and authentication; end-to-end tested with Playwright.
  • Cut dashboard load time from 3–5s to under 1.5s (70% faster) using 25 targeted database indexes (85% query speedup).
  • Modeled a 14–27 day payback for a typical customer; the target market is 300K+ independent U.S. restaurants (a ~$1T supply market).

Custom MCP Server Work | 2025 – Present

  • Servers written from scratch: audio-ears, a ComfyUI image/video server, an NVIDIA-GPU monitor, a YouTube automation server, and an Ollama delegation layer.

YouTube Content Automation Pipeline | Nov 2024 – Present

  • A Python pipeline (PostgreSQL + n8n + Claude API) with a normalized schema, parameterized queries, OAuth2/API-key auth, and webhook workflows; 90% of generated content passes review. Documented end to end as a repeatable build pattern.

Web & Creative Work

Built and shipped front-end sites: this portfolio, a personal site, a site for my father, and the SupplyScout landing page. Music and video production on the side.

Professional Experience

IT Support Analyst | The Boca Raton Resort | Boca Raton, FL | Mar 2026 – Present

Embedded technical support inside a Forbes Five-Star resort's live production systems. The closest day-job analog to forward-deployed, white-glove enterprise support.

  • Front-line support for resort associates and guests across hardware, software, and network; maintain workstations, printers, and point-of-sale systems (Agilysys, Micros, SpaSoft).
  • Manage service requests through Kaseya Autotask; monitor for alerts and outages and escalate complex incidents.
  • Support deployment projects and system upgrades alongside IT leadership; document procedures and produce service-performance reports that drive operational improvements.

Concierge Supervisor | The Boca Raton Resort | Boca Raton, FL | Aug 2025 – Mar 2026

Power user, trainer, and the integration point across 5+ enterprise platforms.

  • One of the concierge supervisor team; led a team of about twelve. Power user and trainer for Opera Cloud, Kipsu, Knowcross, SevenRooms, and Jonas Unify.
  • Trained associates on those platforms and served as a go-to trainer across the enterprise tools.
  • Wrote procedures and a triage workflow for handling and routing guest requests.
  • Coordinated with IT on escalations, getting complex incidents to the right people.

Front Desk Agent | Cloister, The Boca Raton Resort | Boca Raton, FL | Jan 2024 – Aug 2025

  • Used Opera Cloud (PMS) and Kipsu at the front desk for check-ins and guest messaging.
  • Found and reported PMS bugs to IT that fed into resort-wide system improvements; helped test new system rollouts before launch.

Food & Beverage / Retail Customer Service | Earlier

  • More than two decades as a server in food and beverage, with retail customer service (associate and manager) before that. Three decades of front-line customer relationships.

Education

  • Associate in Applied Artificial Intelligence| Miami Dade College | Jan 2025 – Dec 2026 (expected). Coursework: Natural Language Processing · Machine Learning · Neural Networks & Deep Learning · Python for AI · Text Classification & Embeddings.
  • Computer Programming | Broward College | 2004

Certifications (in progress)

  • Anthropic Claude Certification (2026) · CompTIA A+ (Summer 2026) · Google IT Automation with Python (2026) · Google AI Essentials (2026)

Technical Skills

  • AI & agents: Claude Code, MCP servers, multi-agent orchestration, sub-agents and agent skills, LLM application development, Claude API, prompt engineering and structured outputs, RAG / retrieval, vector databases (Qdrant), embeddings, evaluation and confidence scoring, n8n AI nodes, Ollama, ComfyUI, CUDA/TensorRT inference
  • Languages & data: Python, SQL (PostgreSQL, query optimization), JavaScript, JSON, REST APIs
  • Infrastructure & tools: Next.js, Docker, Git/GitHub, Vercel, Supabase, AWS Textract, Tailscale, Qdrant, WSL2, PowerShell/Bash, Windows/Linux, CUDA / GPU inference
  • Enterprise systems (hands-on): Opera Cloud (PMS), Kaseya Autotask, Agilysys/Micros/SpaSoft (POS), Kipsu, Knowcross, SevenRooms

Recognition

  • Employee of the Quarter, The Boca Raton Resort (2025) · Nominated, Platinum Service Awards (Boca Raton Chamber)
  • Internal promotion from hospitality into IT · Bilingual: English (native), Haitian Creole (conversational)

See the work → · Get in touch →