Teaching & Research

The same method.
Different rooms.

Three university courses and one peer-reviewed paper — all built on the same first-principles approach: understand the problem before you reach for a tool.

Courses

Three courses.
One framework.

Undergraduate

Introduction to Artificial Intelligence

A first rigorous contact with AI — from search and logic to production machine learning. Students leave able to formulate a problem, choose a method, and implement it without reaching for a library they can't explain.

  • —Search algorithms — BFS, DFS, A*
  • —Knowledge representation & logic
  • —Supervised & unsupervised learning
  • —Neural networks — from perceptron to backprop
  • —Computer vision fundamentals
  • —Natural language processing
  • —Ethical implications of AI systems
Undergraduate

Data Structures & Algorithms I

The foundations. Students build each structure from scratch and learn to reason about correctness and efficiency before they ever reach for a library.

  • —Arrays, linked lists, stacks, queues
  • —Recursion & divide-and-conquer
  • —Sorting — merge, quick, heap
  • —Binary trees & BSTs
  • —Time and space complexity — Big-O analysis
  • —Hashing & collision resolution
Undergraduate

Data Structures & Algorithms II

Where the first course ends, this one begins. Graph theory, dynamic programming, and computational complexity — the tools that separate engineers who can write code from engineers who can design systems.

  • —Graphs — representation, traversal, shortest paths
  • —Dijkstra, Bellman-Ford, Floyd-Warshall
  • —Minimum spanning trees — Kruskal, Prim
  • —Dynamic programming
  • —Greedy algorithms
  • —NP-completeness & approximation

Published research

From classroom
to production.

Peer-reviewed · UAEH & IPN · México

Deep Learning Applied to Digital Identity Validation

F. Santander Baños · J.H. Sossa Azuela · E. Dorantes Sánchez

A production AI platform for remote identity validation in México integrating document antispoofing and facial liveness detection using an adaptive dual-opinion ensemble strategy. Deployed in production at ~99% accuracy under continuous review.
  • —Adaptive dual-opinion ensemble combining two complementary detection models
  • —End-to-end pipeline: document capture → OCR → antispoofing → face liveness
  • —~99% accuracy in production under continuous adversarial review
  • —ISO/IEC 30107 compliance for Presentation Attack Detection (PAD)
  • —Deployed at national scale — México
Computer VisionFace AntispoofingDocument AntispoofingModel EnsemblesISO/IEC 30107Production AI
Production accuracy
~99%
Under continuous adversarial review
↑Universidad Autónoma del Estado de Hidalgo (UAEH)
↑Instituto Politécnico Nacional (IPN)
Read the paper ↗