M.S. in Data Science student at Fordham University in New York, and a Graduate Research Assistant studying socio-economic indicators of housing distress. I work at the intersection of machine learning, data engineering, and applied research, and I am preparing for doctoral study.
Ongoing research and work in progress.
Analyzing NYC housing data from UNHP alongside Census-tract data to identify correlations between socio-economic indicators and housing distress. Building data pipelines and predictive models to surface insights for a forthcoming paper.
Research and industry experience in data science and software engineering.
Analyzing NYC housing data from the University Neighborhood Housing Program (UNHP) and Census-tract data to identify correlations between socio-economic indicators of housing distress.
Building data processing pipelines and predictive models, and preparing a paper to share findings and insights.
Evaluated large language model (LLM) performance on text generation, code completion, and mathematical reasoning tasks using automated benchmarks and human evaluation.
Worked with exact match and pass@k metrics to assess model accuracy.
Led a team of 9 developers to build a data warehousing platform aggregating data for over 1,000 restaurants.
Automated data population using OCR for physical menus, plus web scraping and the Google Places API for digital ones, using LLMs as a semantic translator to structure data.
Contributed ~40% of the mobile application code using Flutter/Dart.
Built backend services for an AI-powered learning platform that helps graduating students prepare for national exams.
Reduced average analytics query latency by 30% by redesigning PostgreSQL indexes and optimizing Snowflake joins.
Designed a RAG-based AI assistant for Math and Physics with LangChain, improving benchmarked math solution accuracy from 53% to 89%.
Selected academic and personal projects.
Spoken language identification for 9 languages from audio data using CNN and LSTM deep learning models.
Analysis of NYC ride-sharing patterns through supervised and unsupervised learning on TLC trip data.
Shopping-behavior analysis combining dimensionality reduction and density-based clustering to segment customers.
A chess engine using search algorithms and magic bitboards. Implemented Minimax with Alpha-Beta Pruning and Zobrist Hashing to optimize decision-making.
Academic background and relevant coursework.
Tools and technologies I work with across data science and software engineering. More at github.com/biniyam112
Professional training and programs.
Stanford Online
UNDP — UN Frontier Tech Leaders Programme
Graduate of the 2025 United Nations Development Programme Machine Learning cohort.
Africa to Silicon Valley (A2SV)