Tadesse ZeMicheal (347)-(XXX) XXX-XXXX ⋄ XXXX@XXXX.XXX ⋄ linkedin.com/in/tadeze ⋄ github.com/tadeze
EDUCATION ● PhD in Computer Science Oregon State University, Corvallis OR
Adviser: Dr. Thomas Dietterich
Thesis: Anomaly detection and probabilistic diagnosis for automated data quality control ● M.S. Computer Science Oregon State University, Corvallis, OR
Thesis: SENSOR-DX: Machine Learning framework for multi-view anomaly detection and diagnosis ● B.S. Computer Engineering Eritrea Institute of Technology, Asmara Eritrea
SKILLS Languages & Application
Python, C++, R, Java, C#, Pytorch, Tensorflow, SQL, Spark, Mongodb EXPERIENCE
Graduate Research Assistant
Oregon State University Corvallis, OR September 2013 - Present ● Performed research on combining non-parametric anomaly detection with a probabilistic model for identifying
anomalies and diagnosis in time series data.
● Developed Sensor-DX a probabilistic machine learning system for detection and diagnosis of faults in
timeseries. Achieved good detection rate for common faults in weather sensors.
● Developed a mixture of linear models for detection of anomalies in zero-inflated data.
● Developed method for explanation of anomaly detection for fraud and insider threat detection.
Machine Learning Researcher/Engineer TAHMO (Trans-African Hydro-Meteorological Observatory), Jan
2015 – December 2019
● Analyzed data spatial weather data to understand common failures across a network of sensors.
● Developed a flask based ML model to detect anomalies from time series of IoT sensors. Deployed as a
microservice to IBM cloud.
● Mentored and supervised team of developers for ticketing system, visualization systems for weather networks.
Liya enterprise, Asmara Eritrea Aug 2010 - July 2013 ●
Designed and developed enterprise applications for tracking and managing agricultural activities, demographics
and education systems for regional government administration in C# and SQL server. RELEVANT PROJECTS
Anomaly detection using deep stacked autoencoder
Jan 2016 - March 2016 ● Developed deep learning models using stacked autoencoder for anomaly detection. Achieved competitive
performance against the best anomaly detector on anomaly detection of a benchmark dataset.
ADAMS (Anomaly Detection At Multiple Scale)
Sep 2013 - 2015 ● Performed research on anomaly detection and explanation for insider threat detection in a
rganization as p
DARPA funded program.
● Developed an ensemble trees based algorithm for anomaly detection and explanation using C++ with Python
interface. Achieved one of top performing detection among the benchmark algorithms.
Zemicheal, T & Dietterich, T.G., (2019) Anomaly Detection in the Presence of Missing Values for Weather Data Quality
Control ACM COMPASS 2019
Dietterich, T.G., & Zemicheal, T. (2018). Anomaly Detection in the Presence of Missing Values. ODD v5.0, SIGKDD 2018
Workshop, London UK
Zemicheal,T,..etl. (2017). Probabilistic Multi-view based Diagnosis and Anomaly Detection of Sensors in Weather Stations.
poster session presented at workshop NeurIPS 2017 Long Beach, CA.
RELEVANT COURSEWORK Artificial intelligence, Machine learning, Deep learning, Reinforcement Learning & Planning, Probabilistic graphical models
Theory of statistics I-III, Time series analysis, Big data analytics, NLP