You will get a custom AI agent live in production (RAG chatbot to multi-step agent)
Project details
I build AI agents that handle real work in production: answering questions, taking actions, running multi-step processes. Pick the tier that matches your need, or book the simplest and we'll right-size it on the call.
Every tier is built with any major LLM (Claude, OpenAI, Gemini, etc.), deployed to your cloud, documented, and handed over so you or your team can maintain it.
• Simple Agent: a fully custom-coded single-purpose agent, one integration, cloud deployment, docs, a 30-min walkthrough, and 14 days of support.
• Knowledge Agent (RAG): everything above, plus content ingestion (up to 5 sources), a vector database with retrieval tuning, an eval suite for answer quality, and a web UI (embeddable or standalone).
• Multi-Step Agent: everything above, plus architecture design, multi-step orchestration (LangChain, LangGraph, CrewAI, etc.), up to 3 integrations, a full eval and testing suite, and 30 days of support.
Backed by 7+ years across the stack: data engineering at Hewlett Packard Enterprise, ML systems at HelloFresh, and AI-native product work at Enso.
Every tier is built with any major LLM (Claude, OpenAI, Gemini, etc.), deployed to your cloud, documented, and handed over so you or your team can maintain it.
• Simple Agent: a fully custom-coded single-purpose agent, one integration, cloud deployment, docs, a 30-min walkthrough, and 14 days of support.
• Knowledge Agent (RAG): everything above, plus content ingestion (up to 5 sources), a vector database with retrieval tuning, an eval suite for answer quality, and a web UI (embeddable or standalone).
• Multi-Step Agent: everything above, plus architecture design, multi-step orchestration (LangChain, LangGraph, CrewAI, etc.), up to 3 integrations, a full eval and testing suite, and 30 days of support.
Backed by 7+ years across the stack: data engineering at Hewlett Packard Enterprise, ML systems at HelloFresh, and AI-native product work at Enso.
AI Development Type
Deep Learning, Knowledge Representation, Model Tuning, Recommendation System, Software MaintenanceWhat's included
| Service Tiers |
Starter
$897
|
Standard
$1,799
|
Advanced
$2,999
|
|---|---|---|---|
| Delivery Time | 7 days | 14 days | 28 days |
Number of Revisions | 1 | 1 | 2 |
AI Model Integration | |||
Detailed Code Comments | |||
Knowledge Graph | - | - | - |
Model Documentation | |||
Ontology | - | - | - |
Source Code | |||
Taxonomy | - | - | - |
Optional add-ons
You can add these on the next page.
Additional Revision
+$99
Admin dashboard / UI
+$299
30 days extended support
+$129Frequently asked questions
About Alejandro
AI Engineer | Production AI Agents & Automation (ex-HP, HelloFresh)
Berlin, Germany - 1:15 am local time
Most AI projects stall because the engineer can do the ML but not the deployment, or the deployment but not the ML, or both but not the business conversation. I do all three. That's the gap I close for businesses of every size, from solo founders to enterprises.
— RECENT OUTCOMES —
• HelloFresh (global FoodTech): Built two ML APIs (fine-tuned OpenAI CLIP + NVIDIA SegFormer/Meta SAM 2 in PyTorch) that automated image post-production across global brand marketing. Saved €90K+/year. Delivered the full pipeline: training, deployment, monitoring, and integration into the designers' Photoshop workflow.
• HelloFresh (global FoodTech): Rolled out n8n on AWS as the workflow automation platform for marketing teams across Berlin and New York. Set up the infrastructure, trained the teams, and kept it compliant with enterprise data governance.
• Enso (AI-native e-commerce startup): Partnered with the CTO to replace traditional product search with conversational AI discovery. Users could describe what they needed instead of guessing keywords, improving funnel conversion.
• Hewlett-Packard Enterprise (enterprise IT, for Telefónica/telecom): Built the reporting platform that replaced manual weekly reporting with near real-time dashboards for 250+ enterprise clients. Led a small Scrum team and owned the product roadmap.
• Masterschool (online tech academy): Taught Python, SQL, and Data Analytics to 100+ students across bi-weekly live sessions. 97% positive feedback, with several students mentioning my sessions as the reason they enrolled in the full program.
— WHAT I BUILD —
• AI Agents: Autonomous systems built on Claude, OpenAI, and enterprise-grade AI platforms (AWS, GCP). Orchestrated with LangChain, LangGraph, CrewAI, or n8n. Tested and evaluated, not just shipped. From single-purpose chatbots to multi-step workflow agents that handle real business processes end-to-end.
• Workflow Automation: n8n, Airflow, custom Python orchestration. From "automate this one process" to "rebuild our entire ops layer". I'll help you figure out which one your problem actually is.
• Production ML: Fine-tuned LLMs, prompt engineering, RAG systems, computer vision, full MLOps pipelines on Databricks, AWS, and GCP. From research notebook to live API.
• Full-Stack AI Apps: FastAPI backends, React (Next.js) + TypeScript frontends, Postgres + pgvector or Pinecone for vector search, deployed on AWS or GCP with Docker and Kubernetes.
— HOW I WORK —
1. Free 30-minute discovery call. I diagnose what you actually need, which is often less than what you think you need. Sometimes the answer is a $400 n8n workflow, not a $40K custom build. I'll tell you which.
2. Quick prototype in week one. You see something working before committing to the full build.
3. Production delivery with tests, documentation, CI/CD, infrastructure-as-code, and observability. So your team (or the next engineer) can maintain it without me.
4. Optional retainer if you want me available for changes and support. Full handover if you don't. Either way, the system won't depend on me to keep running.
— WHY THIS WORKS —
Most AI freelancers fall into one of two camps: generalist Python devs who picked up LangChain last month, or research-leaning ML engineers who've never deployed anything past a Jupyter notebook. I'm neither. I've shipped production AI inside a company that runs 24/7 across multiple time zones, and the systems I built are still running.
I'm also an educator at heart, which means I won't bury you in jargon. When we talk, you'll understand what's being built and why, not just what it costs. No black-box engineering.
— READY WHEN YOU ARE —
If you have a defined project, send a brief and I'll respond within 24 hours with a plan and a price.
If you don't know what you need yet, that's fine. That's what the discovery call is for. Free, no pitch, no funnel. Just a conversation about what's eating your time and how to give it back.
Steps for completing your project
After purchasing the project, send requirements so Alejandro can start the project.
Delivery time starts when Alejandro receives requirements from you.
Alejandro works on your project following the steps below.
Revisions may occur after the delivery date.
Discovery + architecture
We define the job and success criteria, I design the approach.
Build + deploy
I build, test, and deploy the agent to your cloud.