About.
I am an AI engineer and cybersecurity researcher building systems that protect institutions from a generation of threats they were never designed to detect.
My current research, SentinelVerify, is an AI-powered platform for identifying synthetic identity documents — credentials forged not by hand or printer, but by generative models that have reduced the cost of fabrication to nearly zero. Built in Python with FastAPI and PostgreSQL, it pairs computer vision with machine learning to flag artifacts invisible to legacy verification systems.
The system combines five independent detection signals: metadata forensics (EXIF, C2PA provenance, AI software markers), Error Level Analysis for compression artifact inspection, FFT spectral-domain analysis for diffusion-model fingerprints, transformer-based vision reasoning, and convolutional neural networks fine-tuned on synthetic-image corpora. The aggregation layer is calibrated for ID cards and passports.
I came to this work by way of more than a decade in software engineering. Trained at the Universidad Tecnológica de Panamá, I spent my early career building digital infrastructure for distributors of global brands — most notably JVLAT, the authorized Nintendo and Roku distributor in Latin America, where I led the Magento 2 e-commerce platform and an internal request management system serving multiple countries.
Before that, I built CRM and lead-generation systems at Emporium Developers in Panama, automating workflows that closed apartment sales priced up to $350,000 USD. After relocating to North America, I joined Metropol in Victoria, BC as a Project Manager and Web Developer, where I led UI/UX and database architecture for corporate platforms.
The transition from infrastructure work to AI research wasn't a pivot — it was a continuation. The same problems that plagued e-commerce verification a decade ago have evolved into existential trust problems for entire institutions in 2026. The tools changed; the core question didn't.
Alongside SentinelVerify, I am building socialmaker.ai, a SaaS platform for AI-powered social media content creation — Next.js, React, Prisma, and the Anthropic Claude API at its core. Where SentinelVerify is research-focused, socialmaker.ai is a commercial product designed to test how generative AI can be operationalized responsibly for everyday creative work.
Both projects share a thesis: AI infrastructure is moving faster than the trust frameworks around it. The work I do is on either side of that gap.
What does it mean to verify a person in a world where any face, voice, or document can be synthesized? What infrastructure does a society need to remain trustworthy when the marginal cost of forgery approaches zero? Who is responsible when an AI system mistakes the real for the fake, or the fake for the real?
These aren't abstract questions. They are operational problems that banks, immigration agencies, healthcare systems, and notarial bodies face right now — problems that get worse every quarter as generative models improve. My work is an attempt to be useful on a small, technical part of that surface.
Based in Sandy, Utah. Native Spanish, fluent English. Available for select consulting engagements in AI engineering, cybersecurity research, and document verification systems.
Best ways to reach me: androskent@gmail.com or LinkedIn ↗.