Looking for a job

Oziel Sauceda

@ozielsauceda

Message

I build resilient AI and cybersecurity systems from research to response.

United States

Message

What I'm looking for

I'm looking to build practical AI, cybersecurity, or software systems where I can apply adversarial ML, detection engineering, and full-stack development to solve measurable real-world problems.

At UTRGV's College of Engineering and Computer Science, I develop adversarial reinforcement-learning systems for traffic signal control. My SignalGuard framework models attacker-defender interaction as a Stackelberg game and reduced attack-induced intersection pressure by 59% versus an undefended baseline.

I've built CyberCat, an XDR/SOAR platform that correlates identity and endpoint telemetry into MITRE ATT&CK incidents. It provides traceable detections, real Wazuh response actions with revert support, and near-real-time updates for analysts.

I also built secure multi-operator authentication and RBAC across CyberCat, using OIDC SSO, local accounts, signed sessions, and typed API contracts. My work spans Python, FastAPI, PostgreSQL, Redis, Next.js, and TypeScript.

As a HackResearch 2025 team lead, I helped win a 24-hour hackathon by identifying an anomalous espresso shot from 13 unlabeled brewing logs through time-series feature extraction and distance-based outlier scoring.

Experience

Work history, roles, and key accomplishments

Current

A.I. Research Assistant

Current

UTRGV College of Engineering and Computer Science (MECIS)

Sep 2023 - Present (3 years)

Developed an adversarial RL framework (ATLA) for traffic signal control, formulating attacker-defender interaction as a Stackelberg game within a State-Adversarial MDP. Built a 6-attack evaluation suite and integrated an LSTM into the policy network to improve decision stability under partial observability.

Python PyTorch Sumo Auto LiRPA NumPy Conda Reinforcement Learning Adversarial Machine Learning