AI engineer with a security-first mindset.
I'm Javier Sarmiento. I've spent 3+ years building, deploying, and securing production systems, and I hold a scholarship-funded Master's in Computer Security Engineering and Artificial Intelligence.
- Master's
- MSc Computer Security Engineering & AI
- Universitat Rovira i Virgili · 2025–2026 · Scholarship
- Industry
- Software Engineer, Wazoku
- Enterprise SaaS, AI team · 2023–2025
Experience
Wazoku
Software Engineer · Enterprise SaaS
May 2023 – May 2025 · 2 yrs
On the AI team, built the frontend and backend of an AI-powered agent that automates user interactions on the platform.
- Frontend modernization
- Migrated a critical legacy module from AngularJS to modern Angular, improving maintainability.
- Survey feature redesign
- Led a high-impact product enhancement, reworking the survey feature with the product team.
- Homepage and analytics backends
- Built performant APIs and reusable services for a new homepage, and supported the infrastructure behind client reporting.
Datarte.art
Full-Stack Developer
Feb 2021 – Aug 2022 · 1 yr 6 mos
Built two full-stack apps, Datarte and OtrasManeras, for artist artwork management using React, Gatsby, WordPress, and Flask.
- Cloud infrastructure
- Deployed and managed AWS services (EC2, RDS, S3) for UK clients CityRelay and CityRelaySolutions.
- Data and automation
- Built scrapers, migration tooling, and dashboards using Python, SQL, and Metabase.
- Testing and quality
- Implemented testing with Cypress, Pytest, and Selenium, and helped establish best practices.
Catholic University of Colombia
Research Collaborator
Apr 2020 – Nov 2020 · 8 mos · Colombia
Built a Fitbit smartwatch app streaming heart-rate and calorie data to a web-socket server, for research with Complutense University of Madrid.
Statistics on encrypted health data
Privacy · Cryptography
Homomorphic encryption app (Domingo-Ferrer scheme) that computes statistics on encrypted health data without ever decrypting it.
(opens on GitHub in a new tab)Privacy-Homomorphism-AppPython
Real vs. AI-generated image detection
Machine learning · Computer vision
Classifier that distinguishes real photos from AI-generated (Midjourney) images using feature engineering.
(opens on GitHub in a new tab)identifying-real-images-and-ai-generated-imagesJupyter Notebook
Supervised learning from scratch
Machine learning
(opens on GitHub in a new tab)Supervised learning models built from scratch and benchmarked against library implementations, covering regularization, cross-validation, and ensembling.
Prediction-with-Supervised-Learning-ModelsJupyter Notebook
Crack segmentation and plate recognition
Computer vision
(opens on GitHub in a new tab)Image segmentation and pattern recognition coursework: surface crack detection and license-plate character recognition.
crack_segmentation_taskJupyter Notebook
Metaheuristics for graph coloring
Optimization
(opens on GitHub in a new tab)Genetic algorithm, simulated annealing, and tabu search implementations benchmarked on graph-coloring optimization problems.
optimization_with_genetic_algorithmsJupyter Notebook
MNIST digit classifier
Machine learning
(opens on GitHub in a new tab)Hands-on ML practice, including a digit classifier trained and evaluated on MNIST.
machine_learning_practiceJupyter Notebook
Background
I combine solid full-stack and backend engineering with hands-on machine learning work, LLM-based agents, NLP, and clustering models, shaped by a security-first mindset from academic and professional cybersecurity experience.
I use modern AI tools daily, Claude, Gemini, Cursor, LangChain, to move faster and think clearer, and I'm comfortable owning a problem end to end, from prototype to production.
Currently exploring
- LLM-based agents and emerging protocols like MCP and A2A
- Applied cybersecurity, from a builder's side of the fence
- Investment strategy
- Chess, when there's time for it
Education
Master's Degree in Computer Security Engineering and Artificial Intelligence
Universitat Rovira i Virgili
2025–2026 · Tarragona, ES · Scholarship
Bachelor in Computer Science
Universidad Católica de Colombia
2017–2022 · Colombia
Intensive elite program, top 0.1% of students
Platzi Master
Remote
1:1 training with industry coaches, plus a four-month data engineering project. Strengthened cloud, data manipulation, and software engineering skills.
Skills
- AI & machine learning
- PyTorch · TensorFlow · Scikit-learn · pandas · LangChain · LLM-based agents · MCP / A2A · NLP · Clustering
- Security
- Secure SDLC · Security by design · Vulnerability analysis · Risk management · OWASP Top 10
- Software engineering
- Python · TypeScript · Java · C++ · Django · Flask · Angular · React
- Data, cloud & testing
- PostgreSQL · MongoDB · AWS (EC2, RDS, S3) · Docker · Linux · Git · Pytest · Selenium