Keegan Sanctis | Web Portfolio

Who I Am

About Me

Software Developer & ML Engineer

Software Developer and ML grad student passionate about building impactful full-stack products. Currently pursuing an MS in Applied Machine Learning at the University of Maryland, bridging the gap between software engineering and data-driven solutions.

keegansanctis@gmail.com

Phone

+1 (227) 248-1667

Location

College Park, MD, USA

keegan-sanctis

View Resume

Where I've Worked

Experience

RoyalTech Electrical & IntelliLedge Jul 2024 - Present

Software & Business Developer

Designed and built IntelliLedge, a full-stack SaaS POS and accounting platform using React, Node.js, Express, PostgreSQL.
Implemented role and location based access control, restricting sensitive operations to owners and location specific actions.
Automated data reuse across quotations, delivery orders, and invoices, reducing manual data entry by ~65%.
Developing an ML-based engine to recommend quotation pricing using historical quotation acceptance data.

D2L Sep 2022 - Dec 2022

Software Developer (Co-op)

Kitchener, ON, Canada

Built dropdown and filtering UI components for Brightspace, used by 1,300+ organizations and millions of learners.
Extended a shared UI component library used by 6 teams based on feedback from internal UX stakeholders.
Improved accessibility and usability features across Brightspace, reducing user-reported issues by 15%.

Capco Jan 2022 - Apr 2022

Software Engineering Consulting (Co-op)

Toronto, ON, Canada

Migrated BMO Harris Bank's online banking app from Angular 5 to 13 and Ionic 3 to 5, reducing page load times by 30%.
Developed the login and landing dashboard pages for the banking app, resolving migration-related compatibility issues.
Resolved post-upgrade bugs with QA and maintained a suite of regression tests throughout the framework migration.

Marsh McLennan May 2021 - Aug 2021

Software Developer (Co-op)

Toronto, ON, Canada

Rebuilt the corporate information systems portal with React and Ionic, increasing session duration by 58%.
Implemented monitoring and alerting for Docker/Kubernetes services to help detect system issues and reduce downtime.
Built a CodeceptJS and Playwright regression suite, reducing testing time by 42% and supporting continuous integration.

Pillar To Post Sep 2020 - Dec 2020

IT Support & Quality Assurance (Co-op)

Toronto, ON, Canada

Conducted thorough manual testing, reducing software bugs reported by users by 80%.
Coordinated with cross-functional teams to ensure smooth software deployments, increasing user satisfaction ratings by 50%.

Dye & Durham Jan 2020 - Apr 2020

QA Automation Developer (Co-op)

Toronto, ON, Canada

Developed automated testing scripts using Nightwatch.js for Econveyance's document management software.
Resolved 95% of reported bugs within 24 hours, ensuring robust performance for law firm users.

What I've Built

Projects

Safe Reinforcement Learning for Insulin Dosing

Implementing offline RL algorithms with safety constraints for automated insulin dosing in Type 1 diabetes management. Evaluating conservative policy approaches to ensure clinically safe dose recommendations under uncertainty.

Python PyTorch Reinforcement Learning TD3+BC CQL

In Progress

Portfolio Optimization for Inventory Management

Adapted Markowitz mean-variance optimization to allocate inventory across multiple SKUs using real sales data from RoyalTech Electrical. Formulated inventory as a portfolio problem to balance stock efficiency and demand variability.

Python CVXPY Pandas NumPy Optimization

GitHub

Gun Ownership & Shooting Incident Analysis

Executed the full data science pipeline analyzing the correlation between U.S. gun ownership rates and fatal shooting incidents. Developed and evaluated supervised ML models including Logistic Regression, kNN, and Decision Trees to generate evidence-based policy insights.

Python NumPy Pandas Scikit-Learn Colab

Project

My Background