Michael Negash

Engineered 90% Data Pipelines of OSZ Polls (oszpolls.com): Designed and maintained back-end data pipelines using Node.js to ingest, clean, and process real-time U.S. polling data across 435 congressional districts.

Predictive Modeling & Statistical Analysis: Trained an ML-powered voter model using LASSO regression on 2024 exit polls, generating a dataset of 60,000 synthetic voters to calculate probabilistic party-lean scores Data Simulation & Forecasting — Developed an interactive probability engine that simulates U.S. House election outcomes by analyzing dynamic variables including demographic turnout and party preference

Data Analyst (Data Analytics & Machine Learning) - OSZ Polls

(2025-01)

• Engineered 90% Data Pipelines of OSZ Polls (oszpolls.com): Designed and maintained back-end data pipelines using Node.js to ingest, clean, and process real-time U.S. polling data across 435 congressional districts.
• Predictive Modeling & Statistical Analysis: Trained an ML-powered voter model using LASSO regression on 2024 exit polls, generating a dataset of 60,000 synthetic voters to calculate probabilistic party-lean scores
• Data Simulation & Forecasting — Developed an interactive probability engine that simulates U.S. House election outcomes by analyzing dynamic variables including demographic turnout and party preference
• Data Aggregation & Reporting — Integrated third-party APIs to aggregate live polling data, transforming raw inputs into real-time geographic maps and projected vote share metrics
• Dashboard Development: Built a district-level analytics dashboard featuring interactive visualizations tracking key metrics including polling averages, trend lines, and public sentiment shifts over time
• Advanced Data Visualization: Mapped complex interaction deltas between demographic traits - race, age, and education -using a custom visualization engine to illustrate how multivariate factors jointly impact voting probabilities

Honours Bachelor of Science (HBSc) - Economics - University of Toronto (2021-09 - 2025-05)