I’m Carlos Daniel Jiménez. I’m an AI Software Engineer — I design and build AI systems that work in production. I got here from an unexpected direction: economics, philosophy, and history.
Two lines of work define what I do: MLOps and Agent-to-Agent systems. The first is about making AI reliable at scale. The second is about what happens when the units of computation are autonomous agents talking to each other.
The path here
Currently AI Lead at Nubiral, an AWS partner. I lead a team of AI Engineers, Machine Learning Engineers, and Data Scientists building client products across Banking, Financial Risk, Retail, and Logistics Optimization. Alongside delivery, I drive the adoption of development best practices for AI teams — with a core focus on Agentic workflows and DevOps.
Before that, I worked on problems across very different scales — fraud detection at Mercado Libre, social policy evaluation at the Inter-American Development Bank, edge computing research at Yale with NVIDIA, and ML pipelines at Globant for clients in the UAE and United States.
Earlier: Data Scientist at the United Nations in Chile, and five years at Bogotá’s Education Department building the first AI system that automatically wrote institutional reports.
What I work on
AI Software Engineering — Building AI systems end-to-end: from model design and training to deployment, monitoring, and the software architecture that holds it together in production.
Agent-to-Agent systems — My main research focus. How do agents share context, coordinate decisions, and maintain coherent state across a pipeline? The interesting problems aren’t inside a single model — they’re in the handoffs between them.
Edge Computing — Computer vision on Raspberry Pi and Jetson Nano. ML where cloud isn’t an option.
The longer story
Before I wrote a single line of Python, I spent years studying economics, philosophy, and history — focused on the role of religion in economic development and the doctrines of power in Latin America. Those questions eventually pushed me toward statistics and computer science. Not as a career pivot, but as a natural extension: I needed better tools to answer the same questions.
From there the path follows its own logic. Statistical inference led to machine learning. Machine learning led to MLOps. MLOps, taken seriously, leads to agents. And agents, once you have more than one, lead to the question I spend most of my time on now: how does information flow reliably between autonomous systems?
The thread connecting all of it is the same question I started with: why do systems behave the way they do?
I’ve taught Machine Learning, Applied Statistics, and Bayesian methods across Colombia, Mexico, Peru, Chile, and Spain. I collect vinyl records and build guitars — Jazzmaster-style, because the floating tremolo is the most interesting engineering problem in luthería.
Let’s connect
📬 I write about MLOps, Agentic AI, and Edge Computing. Subscribe to the newsletter →
📍 Based in Colombia, working globally.