Metastatic Cancer Medical Image Analysis Project under Prof. Subha Kalyaanamoorthy
● Achieved 78.28% accuracy for nerve cell image classification using PyTorch by implementing feature engineering and hyperparameter tuning to fine-tune CV Models such as ViT, ResNet, and CNNs, improving baseline accuracy by 4%.
● Visualized and presented findings on model performance to stakeholders using R (ggplot2), Matplotlib, and Pandas.
● Trained models on Compute Canada’s High-Performance Computing (HPC) clusters, managing jobs via Bash scripting.

FHIR-Formatted Healthcare Data Encoding Project under Prof. Bryan Tripp
● Implemented data encoding methods such as Spatial Semantic Pointers and Sinusoidal Encoding in Python to enhance the quality of data inputs for the Transformer Model, improving accuracy by 10.7%.
● Used SQL, NumPy, and SciPy to cleanse, normalize, transform, and structure FHIR-formatted datasets, aiding the ETL pipeline and optimizing the data for model input.
● Utilized Pandas for result aggregation and analysis to generate research insights for team members.

● Automated network traffic detection QA for 20+ applications (Web, Android, iOS) using Python, Selenium, and Appium.
● Analyzed traffic data captured by 60+ application filters by performing packet-level analysis with Wireshark to troubleshoot issues and submit bug reports for 4 filters to improve classification accuracy and enforce proper QoS policies.
● Scheduled recurring automated testing tasks using Linux Crontab to collect testing data.

● Built 50+ web scrapers to contribute to the ETL process for multimedia data collection using Scrapy, Splash, and Lua.
● Executed web scrapers on AWS Batch, monitored their performance using AWS CloudWatch, collected and aggregated document type counts from each scraper post-execution, and leveraged Power BI to visualize document distribution trends.
● Did version control and task management within an Agile Scrum framework using Git, Bitbucket, Jira, and Confluence.

● Analyzed financial data and trends using Excel, identifying inconsistencies in reports to support audit planning.
● Performed detailed testing and data-driven risk assessments to enhance compliance and reporting accuracy.
● Compiled structured insights for audit teams, contributing to the final decision-making process.