Jesse Islam

jesse.islam@mail.mcgill.ca| GitHub| LinkedIn| JesseIslam.com

Visitors so far: Loading…

Summary

Machine learning researcher (PhD, McGill; exp. Fall 2025, available to work now) in causal modeling and time-to-event forecasting. Developed four novel ML methods advancing generative counterfactuals, causal validation, multimodal integration, and survival deep learning, with open-source releases (CRAN, GitHub) and reproducible codebases (Docker, CI/CD, GCP) aligned with publications. Clear communicator (instructor; award-winning TA), with approaches relevant to forecasting, risk modeling, and decision support in finance.

Experience

Machine Learning Researcher (PhD), McGill University

Sep. 2018 – Sept. 2025

Pioneered a counterfactual modeling framework (VAE+GAN) to test “what-if” scenarios on high-dimensional data, improving evidence-based decision-making at scale.
Designed a novel causal network inference method to show how single changes cascade across features toward a goal, pinpointing the most effective levers for outcomes.
Engineered a causal pipeline that cut analysis runtime ∼10× and raised throughput ∼100×, speeding decisions.
Reduced interpretation time ∼70% by merging 10 metrics into a concise 5-tier causal evidence framework.
Prevented $40K+ in wasted research spend by rescuing a stalled project with a novel data-driven framework.
Built an HPC pipeline for large-scale preprocessing, streamlining workflows for stakeholders. Bitbucket
Built a time-to-event ML framework for forecasting event outcomes. GitHub | Publication | Docker
Co-developed casebase (R) for event-rate modeling; published and released on CRAN. CRAN | Publication

myPath Facilitator, McGill

Jan. 2022 – Aug. 2022

Facilitated reflection sessions for interdisciplinary cohorts (groups of 5–10), improving action-value alignment.

Course Instructor, Introduction to Statistical Software (R), McGill

Sep. 2021 – Dec. 2021

Designed and delivered R best-practice lectures; improved comprehension across mixed backgrounds.

Teaching Assistant, Computer Systems, McGill

Sep. 2017 – Dec. 2019

Won student-nominated award (Dec 2019). Simplified complex concepts for 100+ learners.

Selected Publications

Goal-oriented in silico perturbation: a data-agnostic framework using GANs and variational inference for counterfactual analyses. Islam, J. et al. (Expected 2025)
Case-Base Neural Network: Survival analysis with time-varying, higher-order interactions. Islam, J., et al. | Machine Learning with Applications | GitHub (2024)
casebase: An Alternative Framework for Survival Analysis and Comparison of Event Rates. Bhatnagar, S. R., Turgeon, M., Islam, J. et al. | CRAN (2022)

Technical Tools

Programming & Frameworks: Python, R, Bash, PyTorch, scikit-learn, NumPy/pandas, Keras
ML/Modeling: Causality, Time-to-event, Generative AI (VAEs, GANs), LLMs/RAG
Systems & MLOps: Docker, HPC (SLURM), GCP, Terraform, CI/CD (Cloud Build), Git/GitHub

Projects

RAG for arXiv — GitHub
LLM/RAG toolkit to cluster/summarize abstracts; showcases NLP + retrieval and evaluation workflow.
Pivotstage — GitHub
Real-time subject tracking UI using deep-learning-based object detection.
Cloud Resume Challenge — Website | Terraform | Backend | Frontend
GCP app with database APIs, IaC (Terraform) and CI/CD (Cloud Build).

Education

Ph.D. Quantitative Life Sciences, McGill (Montreal, QC) — Expected Fall 2025
B.Sc. Joint Major in Computer Science and Biology, McGill (Montreal, QC) — Sep. 2014 – May 2018