Masoud Karimi
Computer Science @ University of Ottawa
Honours BSc Computer Science (Data Science option) student passionate about building data-driven solutions at the intersection of software development and data science.
Experience
Data Scientist Intern
Ericsson Canada — Radio Applied AI Team
Ottawa, Ontario
May 2026 — Present
- Engineered a modular Python package for automated radio-chart analysis, integrating pretrained ResNet and YOLO models with feature extraction, clustering, and mosaic generation in an object-oriented architecture, enabling interchangeable model backends and reusable analysis workflows.
- Built YAML-based configuration and experiment-tracking infrastructure to define and record model workflows across the pipeline, enabling reproducible experiments and evaluation of new architectures without code changes.
- Contributed radio-chart modeling experiments to an Ericsson-wide Optuna case study, evaluating hyperparameter-optimization reproducibility and practical value to inform its use across internal AI workflows.
- Developed an EVM cloud pipeline on AWS by integrating Amazon S3, Redis, and Amazon EKS, connecting data storage and processing components to support cloud-based EVM workflows.
5G Software Developer Intern
Ericsson Canada — Downlink Physical Layer Test Team
Jan 2025 — Dec 2025
- Designed and implemented a Python-based test log analysis framework that retrieves test data via internal APIs, ingests large scale Layer 1 downlink logs, applies regular expression based parsing, and transforms raw output into structured, queryable JSON artifacts, reducing manual debugging effort by 30% and improving test signal visibility.
- Developed a flaky test comparison tool to identify non deterministic test behavior by diffing multiple test runs using sequence-matching algorithms, enabling rapid isolation of regressions and environment-dependent failures across repeated executions.
- Maintained and refactored Layer 1 downlink test cases by updating Python test scripts and YAML configurations to ensure compatibility with evolving toolchains and execution environments, validating correctness through controlled test execution and result inspection.
- Integrated and validated Python-based automated test workflows within Jenkins CI pipelines and participated in Gerrit-based code reviews, strengthening continuous integration practices and maintaining code quality in a production telecom software stack.
- Collaborated with senior engineers to investigate PHY-layer anomalies, analyze downlink KPIs, and validate transmission scenarios, contributing to improved test coverage and feature stability.
Undergradute Researcher
University of Ottawa, Faculty of Medicine
May 2025 — April 2026
- Developing deep learning models to classify cancer aggressiveness using TCGA patient-derived genomic data, supporting both binary and multi-class (grades 1–4) prediction tasks across multiple cancer types.
- Designed and implemented end-to-end machine learning pipelines including data cleaning, normalization, feature scaling, and train–validation evaluation for high-dimensional biological datasets.
- Tuned neural network hyperparameters using randomized search strategies, achieving a best validation AUC of 0.83 and improving generalization across heterogeneous patient cohorts.
- Applied dimensionality reduction techniques to mitigate feature sparsity and improve model stability when learning from gene-level expression profiles.
- Performed feature explainability and gene-level analysis to interpret model predictions, identifying associations between specific genes and cancer aggressiveness patterns across patient samples.
- Analyzed patient-level trends to investigate correlations between gene expression signatures and aggressiveness across cancer subtypes, contributing biological insight beyond predictive performance.
- Produced reproducible analysis reports and visualizations using Matplotlib, Pandas, and NumPy to communicate results to research collaborators.
Data Analyst Intern
Privy Council Office — Corporate Analytics Team
May 2024 — Aug 2024
- Developed ETL workflows for financial and return-to-office datasets using Tableau Prep and Python, reducing processing time by 20%.
- Automated and optimized data cleaning, transformation, and aggregation pipelines in Python, streamlining multi-source dataset integration and improving data quality.
- Collaborated with internal stakeholders to identify reporting needs and design custom Tableau and Power BI dashboards tailored to departmental requirements.
- Presented insights and dashboard prototypes to cross-functional teams, improving data accessibility and supporting faster, data-driven decision-making.
Data Science Research Assistant
Perkins Bioinformatics Lab — Ottawa Hospital Research Institute
May 2023 — Jun 2024
- Developed a PyTorch logistic regression model for IHEC enhancer data, classifying 1,500 samples into health and cancer categories.
- Performed PCA with Scikit-learn for dimensionality reduction and feature pattern analysis.
- Visualized model performance and PCA results using Matplotlib.
Projects
A collection of projects spanning machine learning, data science, and software development in bioinformatics, finance, and technology.
Filter by Category
Filter by Technology
Soccer Video Analysis
Sports AnalyticsComputer Vision
Computer vision toolkit for analyzing broadcast soccer footage, combining OpenCV and YOLO11 to detect players, classify teams by jersey color, maintain player identities with Kalman tracking, and track the ball. Exports annotated video and frame-level metrics.
Layer 1 Test Log Analyzer
TelecommunicationsSoftware Engineering
Python-based log analysis framework for ingesting large-scale telecom test logs via internal APIs, parsing unstructured Layer 1 output with regular expressions, and transforming results into structured, queryable JSON to accelerate debugging and test signal analysis.
Flaky Test Log Comparison Tool
Test InfrastructureSoftware Engineering
Automated tool for detecting non-deterministic test behavior by comparing multiple executions of telecom test logs using sequence-matching algorithms, enabling rapid identification of regressions and environment-dependent failures.
TCGA Cancer Grade Prediction
BioinformaticsResearch
Transcriptome-based machine learning research using TCGA gene-expression data to predict tumour aggressiveness across 11 cancer types. Compares deep learning models with Random Forest and XGBoost baselines using five-fold cross-validation, gene selection, and feature interpretation.
Get in Touch
I'm always interested in discussing opportunities, collaborations, or just chatting about data science and software development. Feel free to reach out :)
© 2025 · Built with Next.js and Tailwind CSS