Toronto-based AI engineer · ML systems researcher
Mohammadjavad Maheronnaghsh
AI Engineer & Machine Learning Researcher
Agentic AI · Efficient ML Systems · Computer Vision
Toronto, ON · Authorized to work in Canada
I take a top-down approach to AI engineering: start with user impact, define success, establish a
baseline, then build and evaluate. My work spans agentic AI tools, efficient XR systems,
computer vision, and production-minded software.
106x
fewer parameters for Spatons AR predistortion
6x
fewer FLOPs versus CompenNet++ baseline
530 FPS
AR predistortion throughput on RTX 3090
20K+
professor records in Scholar Link outreach platform
3+ yrs
across research, internships, and ML products
Featured work
Systems that connect research quality with real deployment constraints.
Efficient AR
2024 - 2026
Spatons: Efficient Image Predistortion for AR Systems
XR predistortion architecture matching CompenNet++ image quality with 106x fewer
parameters and 6x fewer FLOPs, developed at the University of Toronto with Meta.
- Quality
- SSIM 0.944 / PSNR 27.5 dB
- Efficiency
- 106x fewer parameters, 6x fewer FLOPs
- Deployment
- 530 FPS on RTX 3090; Raspberry Pi benchmarked; Zynq FPGA latency modeled
PyTorch
Model Compression
Edge ML
FPGA
Scholar Link: Agentic Academic-Outreach Platform
Built with 20,000+ professor records, research-area matching, automated personalized
emails and replies, and outreach tracking for graduate-program outreach.
Django
PostgreSQL
Agentic AI
Python
Briefora: Agentic News Briefing
An agentic LLM workflow that turns news into concise briefs, helping readers quickly
understand key developments.
LLMs
Agentic AI
Football Player Detection & Segmentation
Computer vision pipeline for detecting and segmenting football players in sports video.
PyTorch
YOLOv8
U-Net / U2-Net
SAM
OpenCV
Experience
Research depth, engineering execution, and mentoring across applied ML settings.
Jul 2026 - Oct 2026
Arctrade
Artificial Intelligence Engineer
Arctrade
- Supported software quality through AI-assisted code review, testing, and documentation, helping engineers assess changes and identify issues.
- Developed validation workflows that made test coverage and gaps visible to support informed code-change decisions.
- Led refactoring of an internal AI tool to simplify maintenance and support consistent evaluation as requirements evolved.
Sep 2024 - Aug 2026
Toronto, ON
Machine Learning Researcher — ML & XR Systems
University of Toronto · in collaboration with Meta
- Designed Spatons, an XR predistortion architecture matching CompenNet++ image quality with 106x fewer parameters and 6x fewer FLOPs.
- Engineered a geometric lookup-table stage; measured 530 FPS on RTX 3090, benchmarked Raspberry Pi deployment, and modeled Zynq FPGA latency.
- Demonstrated zero-shot transfer to unseen face images without fine-tuning across spatial, tonal, and combined distortions.
- Supervised an undergraduate SystemVerilog implementation and conducted federated-learning experiments for Shannonic.
Jul 2023 - Dec 2023
CISPA Helmholtz Center
Research Intern — Adversarial Robustness
CISPA Helmholtz Center for Information Security
- Implemented and evaluated adversarially robust distillation in PyTorch using a ResNet-18 teacher and MobileNet student on CIFAR-10.
- Studied overfitting and feature-level robustness.
Jul 2023 - Sep 2023
Zista Gene Afarin
Machine Learning Intern — Computer Vision
Biomedical image classification
- Built PyTorch biomedical image-classification pipelines with class balancing and automated workflows for researchers.
- Benchmarked seven transfer-learning backbones, reaching about 98% top-1 and 100% top-3 accuracy.
Feb 2022 - Aug 2026
University of Toronto · Sharif
Teaching Assistant — 20+ Course Offerings
Machine learning, AI, NLP, computer vision, and embedded systems
- Supported more than 20 course offerings, including Lead/Head TA roles.
Research & publications
Research on efficient XR inference, compression, robust vision, and hardware-aware AI.
I work across efficient deep learning, computer vision, edge inference, and hardware-aware
deployment for intelligent systems.
ISCAS 2027 · Manuscript complete
Hardware Accelerator Design for Predistortion Systems for XR Displays
O. Mumtaz · M. Maheronnaghsh · A. Moshovos. Planned submission to ISCAS 2027.
TMLR 2026 · Under review
Spatons: Spatiotonally Localized Building Blocks for Efficient XR Predistortion
M. Maheronnaghsh · S. Kondguli · S. Mann · A. Moshovos.
MLSys 2026
K. Ibrahim · M. Maheronnaghsh · A. Moshovos.
Preprint 2025
M. Hajilou · M. Ansari · M. Maheronnaghsh.
ECCV 2024 OOD-CV Workshop
M. Maheronnaghsh · T. Akbari Alvanagh. Computer Vision – ECCV 2024 Workshops (OOD-CV), pp. 361–379.
Skills
Focused on efficient ML systems, from model design to deployment.
Programming & Systems
Python · C/C++ · C#/.NET · SQL · CUDA · Raspberry Pi · FPGA
GenAI
Agentic AI · LLMs · LiteLLM · AWS Bedrock · OpenRouter · Multimodal AI
ML & Computer Vision
PyTorch · Hugging Face · OpenCV · YOLO · scikit-learn · NumPy · Model optimization
Cloud & Tooling
AWS · CI/CD · Git · Linux · GPU benchmarking
Education
Academic training in ML systems, computer engineering, and applied AI.
2024 - 2026
GPA 3.8 / 4.0
M.A.Sc., Electrical & Computer Engineering
University of Toronto
Supervisor: Prof. Andreas Moshovos · Thesis: Predistortion Acceleration for XR Displays Using Programmable and Highly-Parallelizable Tiny Neural Networks.
Nov 2020 - Jun 2024
GPA 19 / 20
B.Sc., Computer Engineering
Sharif University of Technology
Top 10% · Ranked 106 out of 150,000+ nationally.
