Machine learning scientist

Jesse Islam

PhD-trained machine learning researcher from McGill University developing multi-omics causal-discovery and patient-specific survival-prediction frameworks that boost accuracy, simplify interpretation of large models and slash runtimes tenfold.

About me

I'm an award winning machine learning researcher and teacher, who thrives on tackling complex biomedical data challenges, whether in the lab, the classroom, or AI-genomics competitions.


I've pioneered novel machine learning frameworks using classic statistics and deep learning. Though my training and usecases are biologically/medically driven, I believe in data agnostic methods design and make my code interoperable across different sectors.


If you're curious to dive deeper into my work, you can follow me:

Data Analysis Experience

Machine Learning Researcher
McGill University, Montreal
Sep. 2018 – Sept. 2025

• Pioneered an omic-agnostic VAE+GAN counterfactual framework, unlocking 10 insights per feature.
• Streamlined 10 metrics into a 5-tier causal framework, reducing stakeholder interpretation time by ∼70%.
• Cut in-silico analysis runtime ∼10-fold with an optimized, resource-efficient pipeline.
• Developed omic-agnostic Perturbation Impact Networks to model causal interactions between features.
• Pioneered a deep-learning framework for time-varying effects, dynamically capturing hazard shifts and boosting survival-prediction accuracy.
• Co-developed casebase in R for survival analysis in a team of 3 using version control.
• Extended the counterfactual framework to auxiliary time-to-event outcomes.
• Engineered a deconfounded multi-omics pipeline integrating metabolomics & transcriptomics across tissues to uncover hierarchical insulin response associations.
• Created a custom genpipes gene-expression alignment pipeline to meet stakeholder requirements.

Teaching Experience

myPath Facilitator
McGill University, Montreal
Jan. 2022 – Aug. 2022

• Guided ~20 students through the myPath Individual Development Program in small groups.
• Facilitated reflective discussions aligning research goals with professional growth.
• Coordinated program logistics and tracked participant progress.

Course Instructor — Intro to Statistical Software
McGill University, Montreal
Sep. 2021 – Dec. 2021

• Tailored R-programming lectures to diverse student profiles, boosting comprehension.
• Developed lab materials and assignments to reinforce statistical concepts.
• Held weekly office hours and provided one-on-one support during labs.

Teaching Assistant — Computer Systems
McGill University, Montreal
Sep. 2017 – Dec. 2019

• Empowered 100+ undergraduates by distilling complex circuit-design and assembly concepts.
• Led weekly tutorials and graded assignments with detailed feedback.
• Earned the Computer Science Teaching Assistant Award (Dec. 2019).

Education

Ph.D. Quantitative Life Sciences
McGill University, Montreal, QC
Sep. 2018 – Dec. 2025

B.Sc. Joint Major in Computer Science and Biology
McGill University, Montreal, QC
Sep. 2014 – May 2018

Skills

Key proficiencies that drive my research, analysis, and collaboration.

Machine Learning & Survival Analysis

Design and implement ML models and time-to-event frameworks to predict patient outcomes and boost accuracy by capturing time-varying effects.

Deep Learning & Generative Models

Develop VAE + GAN-based counterfactual frameworks for richer insight into feature perturbations and hazard dynamics.

Causal Inference & Feature Interaction

Build multi-tier causal frameworks and Perturbation Impact Networks to uncover how changes in one biomarker affect others.

Bioinformatics & Multi-omics Integration

Engineer pipelines that integrate transcriptomics and metabolomics data across tissues for hierarchical biological discoveries.

Data Pipeline Engineering

Create optimized, resource-efficient ETL workflows that cut in-silico runtimes by up to tenfold.

Programming & Scripting

Proficient in R, Python, Bash, and SQL for statistical analysis, data manipulation, and automation.

Containerization & Version Control

Leverage Docker and Git to ensure reproducible, collaborative code and research artifacts.

HPC & Workflow Management

Manage large-scale computations with SLURM and cloud-based resources for high-throughput analyses.

Scientific Communication

Publish open-source tools, present at conferences, and mentor students to bridge technical insights with impact.

PivotStage: Video Stabilization Tool

Lightweight Kivy/KivyMD app for automated subject tracking, motion smoothing, and border cropping

Description:

PivotStage uses YOLOv7 to detect and track moving subjects in your videos, applies Savitzky–Golay smoothing for natural camera motion, removes black borders, and re-merges the original audio for a polished final output.

Role:

Architected and implemented core features: integrated YOLOv7 for person detection & tracking, applied Savitzky–Golay smoothing filters, automated border removal, handled audio re-merging, and crafted the GUI with KivyMD.

arXiv-RAG: An arXiv Retrieval & Query Toolkit

Python: arXiv-R toolkit for fetching, embedding, clustering, and querying arXiv papers with RAG

Description:

arXiv-RAG fetches metadata and abstracts from arXiv, embeds them with Ollama’s Llama-based models, applies UMAP + HDBSCAN for clustering, and performs RAG-style queries over your local paper embeddings.

Role:

Designed and built the end-to-end pipeline: handled arXiv API ingestion, integrated Ollama embeddings, applied UMAP dimensionality reduction and HDBSCAN clustering, and enabled cosine-similarity retrieval with RAG query assembly.

CBNN: Case-Base Neural Networks R Package

An R package leveraging Keras for survival analysis with time-varying, higher-order interactions

Description:

CBNN provides R bindings to build case-base neural network models in Keras, enabling flexible smooth-in-time hazard functions and higher-order interaction modeling for survival analysis.

Role:

Developed and packaged the CBNN library: set up Keras integration, implemented end-to-end workflows and authored detailed installation and usage vignettes.

Let's create together

Ready to gain key insights through data? Contact me at jesse.islam@mail.mcgill.ca

Read about some of my works on my blog.

View all posts »

Dive into a realm of design wisdom and creative inspiration, where you'll find invaluable insights, practical tips, and captivating narratives that elevate and enrich your creative journey.