Profile Picture
About Me

Machine Learning Engineer. Scientist

I build machine-learning platforms that turn data into practical solutions. I am currently creating agentic AI tools for healthcare at Peter Munk Cardiac Centre, Toronto General Hospital, Canada.

8

+

Years of Experience

32

+

Projects Done

37

+

Publications

Skills

What I Do

End to end AI and machine learning platforms spanning software development, advanced modeling, multimodal data analysis across text, images, video, graphs, time series, and tabular data, and cloud deployment with MLOps.

  • 01

    Programming & Development

    Python, Kotlin, C++, and SQL development across web, backend, Android, and Wear OS applications, with a focus on scalable and maintainable systems.

  • 02

    Machine Learning & AI

    End-to-end machine learning using PyTorch, Scikit-learn, OpenCV, LangGraph, and AgentKit, covering NLP, large language models, retrieval-augmented generation, computer vision, deep learning, anomaly detection, and statistical modeling.

  • 03

    Model Deployment & MLOps

    Cloud-based deployment and monitoring on Microsoft Azure and Google Cloud Platform using Git, Docker, Flask, and FastAPI, with CI/CD pipelines and containerized model serving.

  • 04

    Data Analysis & System Expertise

    Analysis and integration of graph, image, video, text, time-series, multimodal, and tabular data to support robust AI systems and data-driven decision making.

  • 05

    Research & Scientific Leadership

    Study design, hypothesis-driven modeling, dataset curation, systematic scoping literature reviews, benchmarking and evaluation of machine learning systems, statistical validation, reproducible research, peer-reviewed publishing, grant writing, and research ethics in AI for healthcare.

Resume

Education & Experience

A journey grounded in rigorous training, impactful research, and real-world machine-learning work.

2021 - 2023
Postdoctoral Researcher
@ Institute for Clinical Evaluative Sciences, University of Toronto
2020 - 2022
Postdoctoral Fellow
@ Intelligent Assistive Technology and Systems Lab, University of Toronto
2011 - 2016
Ph.D. in Electrical Engineering & Computer Science
@ Tarbiat Modares University
2007 - 2009
M.Sc. in Electrical Engineering
@ Amirkabir University of Technology
2002 - 2007
B.Sc. in Electrical Engineering
@ Amirkabir University of Technology
2026 - Present
Machine Learning Engineer
@ Peter Munk Cardiac Centre, Toronto General Hospital
2024 - Present
Research Associate
@ Lawrence Bloomberg Faculty of Nursing, University of Toronto
2024 - 2026
Scientific Associate
@ KITE Research Institute, University Health Network
2022 - 2024
Machine Learning Engineer
@ Toronto Rehabilitation Institute, University Health Network
2017 - 2020
Machine Learning Engineer
@ Quark Software
Publications

Publications

A collection of peer-reviewed papers highlighting recent research and collaborations.

  • 01

    OPEN: A Benchmark Dataset and Baseline for Older Adult Patient Engagement Recognition in Virtual Rehabilitation Learning Environments

    Ali Abedi, Sadaf Safa, Tracey Colella, Shehroz Khan

    arXiv · 2025

    https://doi.org/10.48550/arXiv.2507.17959

  • 02

    Artificial Intelligence–Driven Virtual Rehabilitation for People Living in the Community: A Scoping Review

    Ali Abedi, Tracey Colella, Maureen Pakosh, Shehroz Khan

    NPJ Digital Medicine · 2025

    https://doi.org/10.1038/s41746-024-00998-w

  • 03

    Rehabilitation Exercise Quality Assessment Through Supervised Contrastive Learning With Hard and Soft Negatives

    Mark Karlov, Ali Abedi, Shehroz Khan

    Medical & Biological Engineering & Computing · 2025

    https://doi.org/10.1007/s11517-024-03177-x

Blog

Latest Blog Posts

Thoughts and insights on machine learning, AI systems, and real world applications in digital health and beyond.

An interactive series on judging models: confusion matrices, precision and recall, ROC and PR curves, agreement between models, calibration, thresholds, imbalance, confidence intervals, t-tests, test set size, leak-free splits and regression metrics.

Part 1: Reading a Confusion Matrix

Posted on October 12, 2026

Part 1: Reading a Confusion Matrix

A classifier can be right in two ways and wrong in two ways, and a confusion matrix counts all four. Part 1 of the evaluation metrics series walks through the four bins, the threshold that fills them, and what changes past two classes, with three live 3D figures you can turn and tune.

Read more
Part 2: Precision, Recall and F1

Posted on October 12, 2026

Part 2: Precision, Recall and F1

Recall reads across a row. Precision reads down a column. F1 weighs the two together. Part 2 of the evaluation metrics series reads them off the same 100 emails and the same threshold slider, then extends them to more than two classes with macro and micro averaging.

Read more
Part 3: ROC AUC and PR AUC

Posted on October 12, 2026

Part 3: ROC AUC and PR AUC

Each threshold gives one confusion matrix. Sweep every threshold and the matrices trace a curve. Part 3 of the evaluation metrics series builds the ROC and precision-recall curves from the same 100 emails, and shows which area to report and when.

Read more
Part 4: Do Two Models Agree? Bland–Altman Analysis

Posted on October 12, 2026

Part 4: Do Two Models Agree? Bland–Altman Analysis

Correlation is not agreement. Part 4 of the evaluation metrics series compares a spam model with a smaller, cheaper one on the same 100 emails, using the bias between them and their limits of agreement to decide whether one can stand in for the other.

Read more
Part 5: Cohen's Kappa

Posted on October 13, 2026

Part 5: Cohen's Kappa

Two models can agree on most emails just by both saying not spam. Part 5 of the evaluation metrics series counts how often the part 1 model and a cheaper model give the same label, subtracts the agreement chance alone would give, and shows the prevalence trap, how to read the number, and kappa for more than two labels.

Read more
Part 6: Can You Trust a 0.7? Model Calibration

Posted on October 13, 2026

Part 6: Can You Trust a 0.7? Model Calibration

A model is calibrated when its probabilities come true at the rate they claim. Part 6 of the evaluation metrics series draws the reliability diagram for the same 100 emails, scores it with expected calibration error (ECE) and the Brier score, shows why good ranking is not good calibration, and fixes it with Platt scaling.

Read more
Part 7: Choosing the Threshold

Posted on October 13, 2026

Part 7: Choosing the Threshold

Every threshold trades one mistake for the other. Part 7 of the evaluation metrics series puts a price on false positives and false negatives and finds the threshold that costs least, compares it with the rule for calibrated models, Youden's J and the highest F1, and explains why the threshold is chosen on validation data, not the test set.

Read more
Part 8: Metrics for Imbalance

Posted on October 13, 2026

Part 8: Metrics for Imbalance

When spam gets rare, accuracy flatters. Part 8 of the evaluation metrics series makes spam rarer and rarer for the same model and tracks accuracy, balanced accuracy, the Matthews correlation coefficient (MCC) and F1, shows a model that never flags anything scoring 99% accuracy, and explains which question each number answers.

Read more
Part 9: Confidence Intervals and the Bootstrap

Posted on October 13, 2026

Part 9: Confidence Intervals and the Bootstrap

A confidence interval is a range, computed from your data, that is likely to contain the true value of what you measured. Part 9 of the evaluation metrics series defines it and its building blocks, then computes it for the same 100 emails two ways: by resampling the test set 1,000 times (the bootstrap), and with a formula for accuracy, with a Wilson-interval calculator. It also separates test-set uncertainty from training-seed variation, compares two models with a paired bootstrap, and shows how to report each.

Read more
Part 10: The t-Test Family

Posted on October 13, 2026

Part 10: The t-Test Family

Three tests, one idea: a difference divided by its noise. Part 10 of the evaluation metrics series runs the one-sample, paired and two-sample (Welch) t-tests on the series' model scores, shows what goes wrong when paired data gets the two-sample test, and covers the assumptions and the rank-based alternatives.

Read more
Part 11: How Many Test Samples Is Enough?

Posted on October 13, 2026

Part 11: How Many Test Samples Is Enough?

Enough that the reported number is precise enough to make the decision, and no fewer. Part 11 of the evaluation metrics series writes the requirement first, then sizes the test set from the margin of error, from statistical power for a pass-or-fail target, and from the rule of three when no failures are expected, and shows why every subgroup needs its own count.

Read more
Part 12: Splitting Data Without Leaking

Posted on October 13, 2026

Part 12: Splitting Data Without Leaking

Every number in the series assumes the test emails were never seen while the model was built. Part 12 of the evaluation metrics series gives training, validation and test sets one job each, runs k-fold, leave-one-group-out and nested cross-validation, and shows how near-copies from the same sender leak across a random split.

Read more
Part 13: Regression Metrics

Posted on October 12, 2026

Part 13: Regression Metrics

A regression model predicts a number, so every prediction is off by some amount. Part 13 of the evaluation metrics series scores a model estimating heart ejection fraction for 100 patients with bias, MAE, MSE, RMSE, MAPE, R², explained variance, Pearson and Spearman correlation and Lin's concordance, and shows why correlation is not R².

Read more
Get In Touch

Get In Touch

Interested in my research or development projects? Reach out and I’ll get back to you shortly.

LOADING