Production AI Build
Applied AI that holds up in production, not just in a demo.
I design and build applied AI systems end to end: LLM features (RAG, agents, document generation), computer vision, and the data and automation pipelines that feed them. Every build comes with evaluation from day one, a gate that decides what runs on its own and what goes to a person, and the backend, infrastructure and CI/CD to run it.
This is for you if
- You have a clear use case and need it in production, not another prototype.
- The right tool could be an LLM, a vision model or a plain pipeline, and you want it chosen by measuring, not by trend.
- You need AI that works with the systems and data you already run.
- Wrong outputs have a real cost, so results need checking before they reach users or operations.
What you get
- 01
The right approach
LLMs and agents, computer vision or plain automation, chosen by testing on your data.
- 02
Evaluation from day one
Labeled test sets and metrics for the model you use, with an LLM-as-judge calibrated against human ratings where it fits.
- 03
Quality gate
Scores and confidence thresholds decide whether each output is published, revised or sent to a person.
- 04
Production backend
Retries, dead-letter queues, circuit breakers and monitoring, on AWS with Terraform and CI/CD, or on edge devices when the work happens on site.
How it works
01
Intro call
Twenty minutes on the use case, the data and what good looks like.
02
Scope
A written scope with milestones, a timeline and a fixed price.
03
Build
The eval harness comes first, so progress is measured, not guessed. You see working software on real data along the way.
04
Launch
Deployment, documentation and a handover to your team, or ongoing support if you want it.
Pricing
From
US$15,000
Timeline · 4–10 weeks
Fixed price, agreed in writing before work starts.
The starting price fits one focused feature. Larger systems are quoted after the intro call.
After launch: ongoing support (evals, monitoring, fixes) from US$2,500/month.
In production
2 days → 2 hoursAn LLM documentation platform: 2 days → 2 hours per documentIn-house engineer · Fortune 500 manufacturerRead the case study →Questions
Is this only for LLMs?
No. I’ve shipped an LLM documentation platform, a computer-vision system on a Raspberry Pi 4 (published at IEEE CHILECON 2025) and real-time data pipelines. Each build uses whatever fits the problem.
Do you build AI agents?
Yes, when an agent is the right tool. Multi-agent pipelines are part of my production work, and I'll also tell you when a simpler pipeline would be more reliable.
Who owns the code?
You do. The code and infrastructure live in your repositories and cloud accounts.
Can you work alongside our engineers?
Yes. I work in Spanish and English with teams in Mexico and the US, and the handover is built so your team can run the system after launch.
Which cloud do you use?
Most of my production work runs on AWS with Terraform. If you're on another cloud, bring it up on the intro call.
Accepting new projects
Have an LLM feature nobody fully trusts yet?
Twenty minutes is enough to tell whether a Reliability Review or a build makes sense.