My technical capabilities are demonstrated through the complete design, implementation, and ongoing operation of AstroPema.AI — a fully self-hosted production platform deployed on custom-configured Debian/Ubuntu Linux infrastructure. The platform reflects true end-to-end systems integration: hardware selection, RTX GPU optimization for AI workloads, Apache and NGINX web server configuration, PostgreSQL database administration, DKIM-authenticated email services, and real-time deployment of machine learning inference pipelines using Ollama. Every component — from the responsive PHP and CSS front end to the backend AI engines generating interpretations — represents hands-on implementation of production-grade technologies.
Bash scripting is central to my daily workflow. I write, document, and maintain Bash scripts for system automation, log analysis, health monitoring, and security response pipelines. Beyond Bash, I work regularly with Python, PHP, SQL, and standard Linux command-line tooling for system-level tasks and web application support. I understand Debian/Ubuntu package management, service architecture, and system hardening at a demonstrable production level.
My systems run 24/7 with self-built monitoring and incident response. My CNN-GRU intrusion detection system processes Apache logs in real-time, blocking threats with zero false positives in production — a practical example of marrying Bash automation with application-layer security. I also apply Python and Jupyter notebooks to Linux log forensics, turning raw system data into actionable security intelligence. Infrastructure is documented thoroughly — from Bash code comments and system flow diagrams to ISMS-level operational records supporting ISO 27001 certification. Enterprise environments depend on reproducible, auditable processes, not tribal knowledge.
MIT IDSS Machine Learning & Deep Learning | CMU Deep Learning — top 2% both cohorts. BS Mathematics & Computer Science, University of Puerto Rico. 40+ years spanning electronics, telecom, network infrastructure, and enterprise Linux administration. I work independently, manage production infrastructure remotely, and bring the discipline that comes from operating systems where downtime has real consequences.
I hold a Bachelor of Science in Mathematics from the University of Puerto Rico, which provides the analytical foundation for my work in machine learning, data science, and cybersecurity. This background has been reinforced through advanced professional certifications, including the Data Science and Machine Learning certificate from the MIT Institute for Data, Systems, and Society (IDSS) and the Deep Learning professional certificate from Carnegie Mellon University. These programs were mathematically rigorous, competitively graded, and performance-based, emphasizing optimization theory, statistical reasoning, and hands-on implementation rather than tool-level familiarity.
Together, this combination of formal mathematical training, elite technical certification, and long-term hands-on practice demonstrates both theoretical depth and practical competence in designing, deploying, and maintaining secure, scalable AI-driven systems.
- Developed a hybrid CNN–GRU threat detection system with high measured detection performance and low-latency response under typical production load.
- Built a multi-layer defense architecture integrating ModSecurity WAF, Apache mod_evasive rate limiting, and custom ML scoring.
- Reduced backend processing load by aggressive upstream filtering during high-volume attack scenarios, minimizing unnecessary ML evaluation.
- Designed self-hardening pipelines that parse server logs, extract features, score threats, and automate enforcement actions.
- Validated effectiveness against real-world attack traffic patterns (e.g., DNS-over-HTTPS abuse attempts and web shell reconnaissance).
- Operate and harden Linux server stacks for small to mid-sized environments with a focus on reliability and security.
- Manage Apache/Nginx web servers, Postfix/Dovecot mail systems, and PostgreSQL databases, including monitoring and incident response.
- Deploy real-time log monitoring and automated intrusion detection workflows tailored to client risk profiles.
Machine Learning & AI
- CNN–GRU hybrid models
- PyTorch, TensorFlow
- Feature engineering, evaluation
- Real-time inference pipelines
- Reinforcement learning (DQN/PPO)
Cybersecurity
- ModSecurity WAF
- Intrusion detection & response
- Log analysis & alerting
- Rate limiting, abuse prevention
- Hardening & incident triage
Systems Administration
- Ubuntu/RHEL Linux
- Apache/Nginx
- Postfix/Dovecot
- PostgreSQL
- Systemd, iptables/ipset
Programming & Tools
- Python, PHP, JavaScript
- Bash/Shell scripting
- Git version control
- Jupyter, TensorBoard
- API development
- Designed an integrated approach to hardening default Ubuntu/Apache deployments against continuous automated attack traffic.
- Documented practical operational lessons for coordinating multiple security layers (WAF, rate limiting, ML scoring, enforcement).
- Built AI-assisted data workflows for structured dataset generation and model iteration in production-like environments.
- Developed reinforcement learning training pipelines for Atari environments using DQN/PPO architectures.
- Established cost-effective in-house server deployments emphasizing data sovereignty and performance control.