AI products and retrieval
RAG, embeddings, semantic search, agentic workflows, tool calling and MCP/FastMCP integrations.
Evidence: institutional assistants, Spanish retrieval research and tool-using agents.alex@systems:~$ compose backend ai cloud db ready
Backend engineer / Applied AI / Cloud systems
I design the systems behind AI products.
I build the APIs, data layers, retrieval pipelines and cloud infrastructure that turn models into dependable software. From FastAPI and PostgreSQL to RAG, agents, Terraform and production operations.
- Nearly 5 yearsshipping software across industry and public institutions
- Production-mindedmodels, APIs, data and deployment treated as one system
- Technical multiplierreusable tooling and AI workflows for development teams
System capabilities
Four layers. One production mindset.
My best work happens where a model, dataset or operational problem needs to become software that a real team can understand, operate and extend.
Backend and data platforms
Python, FastAPI, PostgreSQL/PostGIS, SQLModel, authentication, file workflows and internal APIs.
Evidence: full-stack platforms, data collection systems and production inference services.Delivery and infrastructure
Docker, Linux, Airflow, AWS, GCP, Terraform, CI/CD, MLflow, DVC and production operations.
Evidence: cloud deployments, ETL operations, MLOps pipelines and DGX infrastructure.AI-enabled engineering teams
Shared agents, reusable skills, project instructions, review flows and structured code generation.
Evidence: standardized delivery workflows used by development teams and technical courses.Frequent tools Python / FastAPI / PostgreSQL / LangChain / MCP / Airflow / Docker / Terraform / AWS / MLflow / PyTorch
Engineering trail
Two accounts, one body of work.
A twelve-month snapshot across personal projects and institutional delivery. Private activity is included only when GitHub exposes it anonymously.
Combined activity
1,447
contributions from 14 Jul 2025 to 13 Jul 2026
@zamax14 ↗- Active days
- 107
- Private
- 315
@alejandroiieg ↗- Active days
- 68
- Private
- 692
- 01 / DataPostgreSQL
- 02 / BackendFastAPI
- 03 / IntelligenceRAG + Agents
- 04 / RuntimeAWS + Linux
Selected systems
Built to survive a technical review.
Products, research and tooling chosen for technical depth and range, not to make the page look busy. Private work is described as a case study without exposing client code.
AulaLLM
Classroom platform with an integrated AI assistant, professor and student roles, content, submissions, grading, file uploads and deployment-ready services.
Architecture Role-scoped relational model, modular FastAPI routers, JWT auth, file attachments, Docker and Nginx.
- React
- TypeScript
- FastAPI
- PostgreSQL
- Docker
Spanish RAG evaluation
Benchmark for Spanish retrieval systems comparing embedding models with FAISS and evaluating ranking quality with Hit@K, MRR and nDCG.
Research path Reproducible evaluation, model comparison and SLURM/DGX execution for thesis experiments.
- RAG
- FAISS
- SentenceTransformers
- Spanish NLP
- Evaluation
Deep Agents Example
Conversational agent with tools, session memory and three interfaces: Chainlit, REST API and CLI.
- Deep Agents
- LangGraph
- FastAPI
- Docker
SyntheticDataLLM
Python CLI that turns documents into Spanish Q&A and reasoning datasets using local language models.
- Ollama
- LangChain
- Docling
- CLI
AI delivery toolkit
Reusable prompts, agents, instructions, hooks and skills for consistent backend, data and testing workflows.
- Copilot
- Agents
- Skills
- Code review
ZMX HomeLab
Docker Compose environment for reverse proxy, monitoring, automation, CI/CD and shared data services.
- Docker Compose
- Nginx
- Jenkins
- PostgreSQL
Experience
The organizations and the progression inside them.
I have moved from embedded software and automation into production ML, data systems, technical leadership and applied generative AI.
IIEG
Institute of Statistical and Geographic Information of Jalisco
Jan 2026 to present
AI Developer
GenAI applications, RAG over documents and databases, FastAPI services, MCP integrations and AI-assisted delivery workflows.
Jun 2025 to Jan 2026
Software Project Leader
Internal software, LLM retrieval systems, ETL coordination and production operations across Linux and cloud environments.
Dec 2024 to Jun 2025
Systems Specialist
Data collection systems, RAG-enabled assistants, OCR and object detection, Airflow pipelines and Superset analytics.
Intelica
Software and AI consulting engagements
Dec 2025 to Mar 2026
Software and Artificial Intelligence Developer
Product-demand forecasting platform on AWS with Terraform, Amazon Aurora, Lambda and serverless frontend hosting; shared coding agents helped the team generate consistent code.
Mar 2024 to Jul 2024
Machine Learning Engineer
FastAPI inference APIs, ONNX Runtime, MLflow, DVC, GitHub Actions and data-labeling pipelines.
Jalisco State Government
Applied machine learning and infrastructure
Machine Learning Engineer
AI processing pipelines, production inference APIs, CNN models and NVIDIA DGX A100 infrastructure.
Bosch
Automotive software engineering
Aug 2022 to Aug 2023
Embedded Software Developer Jr
Automotive production support, software releases, issue resolution and delivery workflows.
Aug 2021 to Aug 2022
Software Automation Intern
Built scripts and automation tools for testing teams while learning industrial requirements, Scrum, Kanban and cross-functional software delivery.
Writing and teaching
If a workflow is worth repeating, it is worth documenting.
I publish practical material when it can help another engineer or a whole team move faster with better context.
CoursePanorama IA LLM
The AI landscape before and after LLMs, including agentic patterns.
Open material ↗ GuideDesign Pattern AgentSpanish documentation and Python examples for GoF design patterns.
Open guide ↗ WorkshopVibe Engineer GuidePractical workflows for shaping coding agents with instructions, prompts and skills.
Open workshop ↗ CourseLocal LLMsFoundations and hands-on local deployment of large language models.
Open course ↗Consulting and collaboration
Bring me the messy version.
A half-defined AI product, a difficult dataset, an internal workflow held together by scripts, or a team trying to build consistently with coding agents. Those are good places to start a conversation.
Available for focused consulting, architecture reviews, technical collaboration and serious conversations about applied AI.
Email Alejandro