Senior AI Engineer · Technical Lead
I'm Rodrigo Díaz Velasco. Since 2021 I've been designing and building products on LLMs, agents and semantic search, working remotely with clients in several countries. Before that, fifteen-plus years in engineering, project management and my own startup.
Selected work
A custom booking engine for a four-hotel chain, synced with its channel manager and connected to an LLM-based virtual assistant. Direct bookings grew by over 50%.
+50%direct bookings
A search platform that understands what shoppers mean. An LLM pipeline processes the catalog, and natural-language queries become structured filters.
A legacy system replaced by a modern platform for workshops, sign-ups, room bookings, payments, cash registers and instructor fees, migrating the old system's data.
A PyTorch pipeline that recognizes tourist landmarks in real time by matching the camera image against a database of descriptors. Exported to ONNX, with a sample Android app.
Read the caseThe specification of an event-driven multi-agent system where nodes pick their own work through topic subscriptions and semantic similarity. It is at the design stage, not implemented yet.
Read the caseMost client code lives in private repositories; I'm happy to walk through it on a call.
What I do
LLM pipelines, RAG and semantic search that work beyond the demo: guardrails, observability, and model choice based on cost, performance and quality.
APIs (FastAPI, Laravel), React frontends and the integrations a business actually needs: channel managers, payments, e-commerce and analytics.
Tests, CI/CD, automated releases and an agent-assisted development workflow that lets me move fast without losing quality.
Twenty years across engineering, international projects and my own startup. I turn business goals into clear technical priorities.
Background
Independent
End-to-end AI products for clients in Argentina, Mexico and beyond, from problem definition to deployment.
Invagrup · FoodTech
Strategy, business development and operations. NAVES finalist (IAE Business School, 2019) and recognized at the APSAL Awards (2020).
Schneider Electric · ABB · Micro Automation · Telefónica
Engineering projects across several countries, with multidisciplinary teams and executive reporting.
Talks & community
The limits of autoregressive LLMs (static memory, no planning, hallucinations), Google DeepMind’s levels of AGI, and the move towards world models: JEPA and DreamerV3.
Talk materialsContact
I'm drawn to problems where AI has to actually work, with real users and real data. Drop me a line and let's talk.
rodrigo@viderlab.com
