Tejas Vyas

Enterprise software architect and computer vision researcher.

Based in Toronto, Canada.

About

I split my time between building production software and researching computer vision. Day to day, I lead architecture and development for a fenestration-industry mobile app suite and ERP system. Alongside that, I run applied AI projects with industry partners at a college AI hub, and I'm completing a PhD focused on computer vision and multimodal models.

I also founded Sotasci (SOTA Science Inc.), where I build local-first, privacy-conscious AI products. You can see that work at sotasci.com. I'm also a co-founder and Principal at GET AI Labs, an applied AI research lab and consulting studio for B2B work.

I like problems that sit at the boundary of research and production: taking something that works in a paper or a notebook and making it reliable, maintainable, and understandable for the people who have to run it.

Experience

Lead Software Architect — Preference North America

Fenestration industry · since 2019

I lead architecture and development for a mobile app suite and ERP system used across the business. Stack: .NET Core, MAUI/Xamarin, React, and T-SQL.

Co-founder & Principal — GET AI Labs

Applied AI research lab and consulting studio

I help shape the lab's AI architecture, research direction, and enterprise system strategy: evaluating emerging methods, designing reliable systems around them, and guiding technical teams toward solutions that hold up in business and public-sector settings. See getailabs.org.

Principal Investigator — Durham College AI Hub

Part-time

I run applied AI projects with industry partners, including 3D reconstruction and camera pose estimation, object detection and segmentation (YOLO/SAM), agentic pipelines, and forecasting.

PhD Student, Computer Science — Toronto Metropolitan University

4th year

My research focuses on computer vision, 3D vision-language models, and compact multimodal models.

Education

MSc, Computer Science (AI) — Toronto Metropolitan University
BSc, Computer Science & BA, Mathematics — University at Buffalo

Research

My current research is in computer vision and multimodal models: 3D vision-language understanding, and making multimodal models smaller and more efficient without giving up capability.

3D vision-language models
Compact, efficient multimodal models
Applied computer vision: 3D reconstruction, camera pose, detection and segmentation

Earlier in my research career, I worked in medical AI, including outcome prediction for mitral valve transcatheter edge-to-edge repair and hypoglossal nerve stimulation therapy.

A full publication list is on my Google Scholar profile.

Work with me

I take on a limited number of engagements in three areas:

For larger applied-AI research programs, from opportunity mapping to deployable prototypes, I work through GET AI Labs.

Fractional CTO & architecture review

Hands-on architecture and technical leadership for small and mid-size companies — reviewing existing systems, planning what to build next, and helping teams ship it.
Applied CV & LLM consulting

Practical computer vision and LLM work, with a focus on on-device and local-first AI, and audits of fine-tuning approaches for teams adopting these models.
Mentorship

Mentorship for graduate students and early-career engineers navigating research, industry, or the space between the two.

Contact

The best way to reach me is by email.