México — AI Lab
Hi. I’m Estefanía León.
AI Engineer.
Building production LLM systems and AI architecture at Natura & Co, and researching representation learning for collider data as an AI Research Intern with CERN’s Next Generation Triggers.
- Machine Learning/
- LLMOps/
- AI Architecture/
- Research
A designer’s eye, an engineer’s hands.
- 01Game Design
- 02Software
- 03Machine Learning
- 04LLMOps
I didn’t arrive at machine learning through a straight line. I started with a degree in video game design and interactive content — learning how structure, feedback, and interface decisions decide whether a system actually works for the person using it. That instinct for structure is what later pulled me toward representation learning: wanting to see the actual shape of things, not just their surface.
The bridge was research. While building web and VR projects, I was also analyzing CERN ALICE open data and training deep learning models to classify particle interactions — my first real exposure to a field that holds models to a standard most industry work never has to meet. Two master’s degrees later (IoT & AI, then Big Data), that research habit turned into a full engineering discipline.
Today I’m an LLMOps Engineer at Natura & Co, designing production conversational AI systems and the architecture that keeps them fast and affordable at scale. In parallel, I’m an AI Research Intern with CERN openlab’s Next Generation Triggers initiative — a focused, fixed-term internship, not the whole story. Outside of all that: Hyrox training, and finding reasons to travel.
A career built out of order, on purpose.
Every stop taught the next one something. The timeline below is chronological — the only place on this site where numbering carries meaning.
- 2021
B.S. Video Game Design & Interactive Content
Universidad Kino
Started in interactive and visual design — learning how structure, feedback, and interface decisions shape whether a system is actually usable. That instinct for structure carried straight into machine learning later on.
DesignInteractive SystemsC# - 2021–2023
Web Developer & VR Creator
CONACYT
Built accessible, responsive government websites and an interactive Oculus-based VR laboratory experience — the work that came right before moving into scientific research.
VRWeb DevelopmentUnity 3D - 2022–2024
Scientific Researcher & ML Developer
UNISON
Analyzed ALICE experiment open data and built ML/deep learning models for hadronic particle detection and classification, developing and optimizing analysis code for the ALICE3 project and collaborating with physicists and data scientists on publication-oriented research.
Machine LearningALICE Open DataScientific Computing - 2024
M.S. Engineering in IoT and Artificial Intelligence
Universidad de Sonora
Deepened the applied ML foundation — moving from research scripts to systems: model deployment, embedded/IoT constraints, and production-minded engineering.
M.S.IoTApplied AI - 2025
LLMOps Engineer
Natura & Co
Designing and shipping production conversational AI systems, leading token-usage and inference-cost reduction through prompt redesign, caching, and model-routing, and contributing to scalable AI architecture and workflow automation.
LLMOpsAI ArchitectureCost Optimization - Jun–Sep 2026
AI Research Intern — Next Generation Triggers
CERN Next Generation Triggers (research internship)
A focused, fixed-term research internship: building PyTorch representation-learning models for LHC Level-1 trigger event embeddings on updated Collide-2V L1-only particle features, evaluating Linformer and multi-head attention architectures for collider-event classification and anomaly detection (HH4B vs. Standard Model backgrounds), and running latent-space diagnostics — PCA, effective rank, Mahalanobis scores, Fisher analysis — with reproducible workflows tracked in MLflow and GitLab.
Representation LearningLinformerLatent-Space Diagnostics - 2026
M.S. Big Data
Universidad Autónoma de Guadalajara
Rounding out the data and ML foundation with large-scale data engineering and analytics — the infrastructure side of building systems that hold up outside a notebook.
M.S.Big DataData Engineering
What I actually spend my time thinking about.
5 threads that keep reappearing across projects, papers, and late-night reading. Open one to go deeper.
Selected work, end to end.
Each project page walks through the problem, the approach, and what actually happened when it shipped.
On explaining this stuff out loud.
Conversations about the path from design to AI, and about what it's like building ML for a physics experiment.