Fsehaye Medhanie

About Fsehaye Medhanie

I build machine learning systems that move from experimentation to real-world impact.With 7+ years of experience across data science, machine learning engineering, and database systems, I specialize in designing production-grade ML solutions that solve measurable business problems. My work sits at the intersection of modeling, infrastructure, and product execution.At Infinity Trucker, I architected and deployed a real-time load recommender system serving daily drivers. By combining predictive modeling, market forecasting, and real-time data ingestion, we reduced recommendation latency by 30% and increased driver profit by 15% in targeted regions. I also implemented monitoring pipelines using Prometheus and Grafana, reducing critical model failure detection time by 90%.My expertise includes:Recommender systemsTime-series forecastingDriver churn modelingA/B testing & experimentationLarge Language Models (LLMs) and RAG systemsMLOps (Docker, Kubernetes, CI/CD, MLflow)AWS SageMaker deploymentI began my career in analytics and database engineering, where I learned how data infrastructure and statistical modeling intersect. That foundation allows me to think beyond models — I design scalable systems.Currently based in Oakland, CA and fully authorized to work in the U.S, I’m interested in Machine Learning Engineer, Applied Scientist, or Senior Data Scientist roles focused on high-impact ML systems in logistics, marketplaces, fintech, or AI-driven products.

Experience

1. Machine Learning Engineer

   Infinity Trucker

   Jan 2024 — Present · TX, US

   Architected and deployed a real-time load recommender system on Kubernetes supporting daily active drivers.Reduced recommendation latency by 30% through optimized model serving and API design.Built predictive market analysis framework using time-series models to identify high-profit regions, increasing average driver earnings by 15%.Implemented model monitoring with Prometheus and Grafana, cutting time-to-detection of production failures by 90%.Led end-to-end ML lifecycle automation using Docker, CI/CD pipelines, and MLflow.Integrated LLM-based retrieval-augmented generation (RAG) to enhance load matching and route decision support.

Education

• UiT- The Arctic University of Norway

  Exchange Program, Big Data and Machine Learning

• Eritrea Institute of Technology

  Bachelor's degree, Mathematics and Computer Science

• ITMO University

  Master of Science (MSc), Big Data and Machine Learning

• Kazan State University

  Preparatory classes , Russian Language

• San Francisco Bay University

  Master of Science - MS, Computer Science

  2025