Featured Projects

The World’s Flagship Home Vision Care App—Born from University of Toronto, Built for the Globe
EyeBuddy began as a student-led innovation inside the University of Toronto’s entrepreneurship lab, with a bold mission: to bring accurate, accessible vision screening and eye care tools to every home in the world. Today, it has grown into our flagship global platform, built by a diverse team of engineers, clinicians, and designers across continents—all focused on redefining how the world checks and protects its vision.

At its core, the EyeBuddy app turns any smartphone into a personal eye health hub. It offers a comprehensive suite of clinically-informed, AI-assisted tests including visual acuity, astigmatism screening, Amsler grid for macular degeneration, contrast sensitivity, red desaturation, color blindness, and eye motility tracking. Users can test their vision in minutes, anywhere, without specialized equipment. But EyeBuddy goes far beyond screening. It features smart medication reminders tailored to ophthalmic regimens, and a growing library of therapeutic eye exercises for dry eye relief, lazy eye (amblyopia), convergence insufficiency, and digital eye strain—each designed in consultation with optometrists and physiologists, then iterated through real-world user testing.

With encrypted health records, geolocation-based specialist matching, and seamless referrals, EyeBuddy doesn’t just test—it connects users to care. Our back-end is built on scalable cloud infrastructure with secure data handling, machine learning feedback loops, and cross-device syncing, ensuring a seamless experience across platforms.

What started as a student idea is now a global vision platform empowering people from Dhaka to Toronto, Nairobi to London—bridging gaps in care and helping detect vision loss early, before it’s too late. EyeBuddy isn’t just an app—it’s a movement toward proactive, tech-enabled eye health for everyone.

Transforming Eye Care at Bangladesh's Largest Eye Hospital
EyeBuddy proudly partnered with Bangladesh Eye Hospital Dhanmondi—the largest eye care institution in the country with over 140 ophthalmologists, 11 branches, and more than 20,000 monthly patients—to build a next-generation teleophthalmology platform. Over 18 months, our team worked closely with hospital leadership, clinicians, and patients to deeply understand the unique demands of the Bangladeshi healthcare landscape.

The result is a fully integrated, highly secure app featuring home-based vision testing using EyeBuddy’s proprietary EPOLLE system, seamless teleconsultation access to any ophthalmologist, and centralized, encrypted medical record storage. Built with a robust backend architecture including scalable cloud systems, advanced encryption protocols, and real-time EHR syncing, the app has redefined digital eye care in Bangladesh.

Since its launch, the EyeBuddy platform has transformed how patients across the country access, manage, and trust their eye care—ushering in a new era of remote ophthalmology.

State-of-the-Art Diabetic Retinopathy Detection Engine Co-Built with India’s Leading Eye Hospitals
EyeBuddy partnered with Lions Eye Institute and Hospital (Chattogram) and Susrut Eye Foundation & Research Centre (India) to co-develop RetinaScan AI—a sophisticated, clinically trained artificial intelligence system for diabetic retinopathy (DR) detection, built to meet the scale and complexity of real-world screening. The project began with a dataset of over 60,000 unlabeled 45-degree color fundus photographs. To create ground truth, we orchestrated a meticulous data annotation process: trained graders and retina consultants manually labeled each image not only for DR stage but also for individual pathological features like microaneurysms, dot-blot hemorrhages, hard and soft exudates, IRMAs, and neovascularization—using a structured feature-mapping protocol. Each image took 20+ minutes to annotate, resulting in one of the most detailed DR datasets in South Asia.To ensure our engineering team had domain-specific insight, we conducted clinical deep dives and cross-disciplinary workshops, training them in retinal anatomy, DR grading scales, and lesion morphology. With this foundation, we engineered RetinaScan AI using an ensemble architecture of convolutional neural networks (CNNs), primarily based on DenseNet-201 and ResNet-50 backbones, integrated with attention mechanisms and multi-task learning branches to detect both global DR grades and local lesion patterns simultaneously. The model was trained and optimized using TensorFlow 2.0 on a custom-built pipeline in Google Colab Pro+, leveraging GPU acceleration, cyclic learning rates, batch normalization, and advanced image augmentation techniques (CLAHE, affine transforms, and color constancy correction) to boost generalizability and reduce overfitting.Post-development, we deployed RetinaScan AI in two formats: (1) a cloud-based API hosted on a HIPAA-compliant secure server stack with real-time inference capabilities and auto-scaling infrastructure, and (2) a lightweight desktop software designed for offline DR screening in low-bandwidth rural environments, using integrated SQLite databases and GPU-accelerated inference for near-instant analysis. Both versions include modular interfaces for clinician feedback, heatmap-based lesion localization (Grad-CAM), and compatibility with major fundus cameras. Our ophthalmologists were embedded in every step—from model validation and ground truth curation to deployment testing—to ensure medical integrity and trustworthiness.We are now conducting pilot studies in live DR screening eye camps across Bangladesh and India, where RetinaScan AI is powering frontline screening with high-speed grading, referral flagging, and lesion-based progression tracking—redefining what’s possible in AI-driven retinal care.

Smartphone-Based Cataract Detection for the Last Mile of Vision Care
EyeBuddy partnered with the Lions Eye Hospital and its network of cataract surgeons to develop CataSee AI—an advanced, smartphone-powered cataract detection system designed to make cataract screening possible anywhere, no slit lamp required. Over two years, we worked hand-in-hand with the surgical teams and patients to collect and label over 40,000 real-world anterior segment images, covering a full spectrum from clear lenses to immature and hypermature cataracts. Our mission was clear: to create an AI that could instantly tell, from a single smartphone photo, whether a person has a normal eye, an early (immature) cataract, or a mature cataract—each with its own set of risks, including increasing surgical complexity and higher complication rates if left undiagnosed.

The model behind CataSee AI uses a hybrid CNN-ViT (Vision Transformer) architecture, pretrained on global anterior segment datasets and then fine-tuned on our locally collected images. We integrated deep feature extraction layers trained to distinguish lens opacities, cortical spokes, nuclear sclerosis, and posterior subcapsular changes, even in low-light and variable environments. The model was trained and deployed using TensorFlow Lite for ultra-fast mobile inference and includes real-time image quality feedback, noise reduction, and explainable heatmap outputs—making it one of the most field-ready cataract detection tools ever built.

Today, CataSee AI is being piloted across rural Bangladesh through our mobile eye camps, where local technicians use only a smartphone to scan hundreds of eyes per day—instantly identifying those needing cataract surgery and fast-tracking them to care. From remote villages to river islands, this tool is helping close the cataract blindness gap and ensuring that no one is left behind simply because they live too far from a hospital. CataSee is more than an AI—it’s a mobile frontline against preventable blindness.

Reimagining Pediatric Vision Screening with CNIB
At the height of the COVID-19 pandemic, screen time surged—and so did concerns about children’s eye health. In response, EyeBuddy teamed up with the Canadian National Institute for the Blind (CNIB) to build a next-generation pediatric vision screening and training app tailored for home use.

Over two years of collaboration, our team worked closely with CNIB’s pediatric vision experts and conducted iterative user testing across Canadian households to ensure the platform was clinically sound and genuinely kid-friendly. The result is a highly specialized app that brings together robust vision testing and playful engagement in a single ecosystem.

Core Features:

Visual acuity, color vision, strabismus, eye motility, and astigmatism tests adapted for home use
A fun suite of AI-assisted games and interactive eye exercises to support amblyopia therapy and relieve dry eyes
Smart child profile tracking with encrypted data storage and progress analytics
Milestone-based gamification and national leaderboards to keep kids motivated
UX design guided by behavioral psychology and validated through CNIB-led field studies
Built using dynamic UI components, real-time visual tracking, and proprietary pediatric optometric algorithms, the app has become a gold standard in remote vision screening and training for children. Today, it's not just a tool—it's a digital companion in every child’s eye care journey.

GlaucoMind AI is EyeBuddy’s upcoming multi-modal system for early glaucoma detection and risk prediction. It combines OCT, fundus, visual fields, IOP trends, and patient history using attention-based CNNs and RNNs. The tool aims to detect structural-functional dissociation before vision loss begins. Cloud and device-compatible, it enables real-time AI analysis from clinic or home. GlaucoMind will empower early referrals, long-term monitoring, and proactive care through the EyeBuddy platform.

KeraSight AI is EyeBuddy’s upcoming tool for early keratoconus detection and progression prediction. Using DenseNet-ViT architecture and LSTM-based temporal analysis, it analyzes corneal maps and biomechanics to flag subclinical changes and forecast risk. It supports Scheimpflug, Placido, and even smartphone imaging. The model is cloud-based and API-integrated for clinics. Designed with cornea experts, KeraSight aims to enable mass screening, early CXL referral, and smarter surgical decisions.