Mercedes-Benz
Smartbox
Nexook
Prozeso
Full-stack runs deep.
Frontend to infra.
Scroll to explore the stack.
Senior software engineer with nearly 8 years shipping production systems across enterprise (Mercedes-Benz), scale-ups (Smartbox), and startups. Hands-on with AI/ML in production since 2022. Currently CTO & co-founder at Prozeso.
Selected work
● ai · e-commerce 2024–2025
Voice shopping assistant
for Smartbox
Led the introduction of AI into a high-traffic e-commerce platform: shipped a voice-based assistant that guides customers through the catalogue acting as a virtual shop assistant, plus a personalized recommendation agent based on purchase history.
Vue 3 Nuxt LLM agents Voice AI Recommendations
Read case study → ● ml · enterprise 2022–2023
DDX Marketplace
Strategic Big Data & ML initiative at Mercedes-Benz where data pipelines and cloud infrastructure were the core priority. Designed microservices and refactored backend for performance.
React NestJS PostgreSQL ML pipelines
Read case study → ● rag · automation 2025–present
Prozeso AI Stack
Production automation infrastructure delivered to 15+ companies. RAG pipelines over private knowledge bases, conversational agents, and LLM-driven workflows across sales, support, ops, and HR.
OpenAI Anthropic pgvector MCP
Read case study → ● self-hosted · personal 2025
Otto
Self-hosted personal AI assistant on legacy hardware. Local LLM runtime, RAG over personal knowledge base, vector storage, and custom MCP integrations. Full ownership of the stack.
Local LLM RAG MCP Self-hosted
View on GitHub → — descending into the stack —
sys.profile
Layer 02 · Backend & Systems
I started writing software in 2018. Eight years in, I've shipped on enterprise teams (Mercedes-Benz), scale-ups (Smartbox in Dublin), and now my own consultancy.
I care about systems that ship, scale, and don't break, and about being the senior who can look at a vague problem and turn it into something running in production by Friday.
I build production AI systems for companies that need to ship, not prototype. My work spans the whole stack: data, infra, model orchestration, and the UI on top.
8y
software engineering
15+
ai systems shipped
04y
ai/ml in production
02x
languages · es / en
// ai & llms
OpenAI · Anthropic · Vercel AI SDK · MCP · ElevenLabs
// data & rag
pgvector · Pinecone · Qdrant · Supabase · PostgreSQL
// frontend
Next.js · React · Vue 3 / Nuxt · TailwindCSS · TypeScript
// backend & infra
Node.js / NestJS · Python / FastAPI · AWS · Docker · Vercel
— going deeper —
root@algoro:~$ ls -la /stack/infrastructure
The layer nobody sees. The one that matters most.
● containerization Docker · AWS · multi-env
● cloud-deployment Vercel · AWS · CI/CD pipelines
● vector-storage pgvector · Qdrant · embeddings
● self-hosted-llm Ollama · local runtime · legacy hw
● data-pipelines ingestion · ETL · RAG indexing
● api-design REST · NestJS · FastAPI · typed
● SELF-HOSTED · RUNNING
Otto
Personal AI assistant running on legacy hardware. Local LLM runtime via Ollama, RAG over my personal knowledge base, custom MCP integrations, and full vector storage — without sending a single token to the cloud.
→ /work/otto — case study— signal detected at depth 04 —
Layer 04 · AI / LLMs / RAG
This is where it thinks.
You've scrolled through the full stack. This is the bottom. A working RAG system over my CV, case studies, and writing — same architecture I ship to clients. Ask anything.
sys.log
- → Gemini 2.5 Flash · streaming
- → Gemini embeddings · gemini-embedding-001
- → In-memory cosine · top-5 retrieval
- → Inline [N] citations · linked sources
- → Vercel · on-demand Astro route
Ask anything about Alejandro's work, projects, or experience. Answers stream live and cite the source.
— end of stack —
You made it to the bottom.
Most people don't.
Open to senior contract engagements and full-time roles at AI-first companies. If you've read this far, you already know if we're a fit.