Denel John — Full Stack+ AI generalist

Attend.io | Facial Recognition Attendance System

Built a college-wide attendance portal that replaces manual roll call and paper sign-in sheets entirely — students check in by scanning their face on their phone at the start of class, and the system automatically marks them present within seconds. Eliminated the risk of proxy attendance (one student signing in for another) by verifying each student's face against their own enrolled photo and confirming they are physically on campus before accepting a check-in.
Faculty no longer take attendance themselves, they simply open a session at the start of class and get a live, accurate register automatically, with the ability to correct any mistakes. Built dashboards that surface attendance trends, flag students falling below 75% attendance, and alert faculty to repeated absences early in the semester.
Admins manage rosters, accounts, and reports from a single portal, with full import and export support.

Leaflytics | Full-Stack AI Plant Diagnostics

Leaflytics is a crop disease detection app that lets farmers and gardeners identify plant diseases instantly by uploading a photo of any affected leaf. Under the hood, it runs a custom-trained EfficientNetV2 model capable of diagnosing 38 plant-disease pairs with 99.13% accuracy on the validation set. The model is paired with a treatment recommendation layer that maps each diagnosis to actionable next steps, so the output is immediately useful, not just a classification label. The goal was to bring specialist-level diagnostic confidence to anyone with a smartphone, without needing agronomist access or lab turnaround time.

TensorFlow/Keras | NestJS | Python | React | Tailwind CSS

Synapse | Visual StudyBoard

Synapse is an AI-powered visual learning workspace that transforms any topic into an interactive study canvas. Instead of chatting in a linear conversation, users build knowledge through connected learning clusters containing deep explanations, ELI5 summaries, visual references, videos, quizzes, notes, and AI-generated study aids, all organized spatially for better understanding and retention.

Shrike | AI video threat detection to instant voice dispatch.

In high-security facilities, traditional CCTV relies on operators manually watching screens and frantically calling field guards during emergencies, wasting critical seconds when lives are at risk.
Shrike solves this by converting passive monitoring into an active, automated response pipeline; it routes live video streams or camera frames through an edge-deployed computer vision model to automatically detect violent incidents, updates an event-driven Supabase Postgres backend, and instantly triggers a loud emergency ringtone and a low-latency, automated Vapi.ai voice dispatch call on a field guard's smartphone handset while streaming live transcripts back to a high-density, tactical operator dashboard.