FRACTIONAL FORWARD DEPLOYED ENGINEER
Español →I DEPLOY LLM AGENTS
inside your business.
Your fractional Forward Deployed Engineer. I deploy, integrate and operate LLM agents inside your company — cloud, on-prem or hybrid. Not a chatbot demo: agents wired into your real systems, running in production, that I keep running.
The technical firepower of a full-time deployment engineer, without the full-time headcount. Month to month, or by project.
Gabriel Neuman
The tools I deploy with
Claude
Next.js
VercelTWO WAYS TO WORK WITH ME
Fractional · month to month
I run your agent stack
I embed part-time as your Forward Deployed Engineer. Deploy, integrate, debug and operate your LLM agents as an ongoing capability — the way a fractional CTO runs your engineering, scoped to agents and automation.
By project · fixed scope
I ship one deployment
A defined outcome: get a specific agent deployed, integrated with your systems and running in production. Fixed scope, fixed price, and it keeps working after handoff.
WHAT I ACTUALLY DO
The work of making agents run inside a real, complex environment — every step shipped, not theorized.
Deploy agents in your environment
Hermes-style LLM agents on Docker across cloud, on-prem or hybrid. Self-hosted inference (vLLM) when the data has to stay inside your walls.
Wire them into your systems
Agents that read and write your real stack — internal REST APIs, databases, CRM, email, messaging. They trigger business actions, not demos.
Debug across every layer
When an agent misbehaves in production, I trace it across infra, orchestration and the model layer. Multi-agent orchestration, cost and observability included.
Operate and iterate
Deployment reliability, observability, and the operational tooling to keep a fleet of agents running. Scoping, build, and iteration against your real environment.
THE STACK I DEPLOY
Agents
Hermes Agent, Claude Code, custom multi-agent — multi-model, no lock-in
Orchestration
Gastown, Trigger.dev, parallel sub-agents, scheduled runs
Infra
Docker, Linux/VPS, Vercel, self-hosted inference (vLLM)
Integrations
REST APIs, MongoDB, Airtable, CRM, email, Slack, Telegram/Discord/WhatsApp gateways
Observability
Agent spend tracking, skill/memory systems, portable skill libraries
App layer
Next.js, TypeScript, MongoDB/Supabase — customer-facing web on the agent backend
EVIDENCE — SHIPPED & PUBLIC
Deployed Hermes Agent in production →
Full Docker/VPS deployment: messaging gateway, natural-language cron, learning-loop skills.
Hermes Agent technical overview →
Architecture, multi-model routing, parallel sub-agents, migration paths.
Client delivery track record →
Production systems for real clients, each with problem, system, stack and measured result.
Full technical profile (llms.txt) →
Machine-readable profile of exactly what I deploy, integrate and operate.