I'm a hands-on machine learning engineer who loves turning messy data into actionable insights. As a team lead at iCog Labs and SingularityNET, I enjoy shaping research directions and building scalable AI systems that perform in real-world settings. My work focuses on efficient model architectures, robust evaluation, and pragmatic tooling to reduce energy consumption and training time while preserving model quality and interpretability. Earlier in my career, I built a full-stack web tool and gained experience across the software stack, from Laravel-based backends to admin dashboards. I thrive in collaborative environments, bridging research and production applications, and I’m passionate about delivering impact through data-driven solutions.

Beza Wit Abebaw Tekeba

I'm a hands-on machine learning engineer who loves turning messy data into actionable insights. As a team lead at iCog Labs and SingularityNET, I enjoy shaping research directions and building scalable AI systems that perform in real-world settings. My work focuses on efficient model architectures, robust evaluation, and pragmatic tooling to reduce energy consumption and training time while preserving model quality and interpretability. Earlier in my career, I built a full-stack web tool and gained experience across the software stack, from Laravel-based backends to admin dashboards. I thrive in collaborative environments, bridging research and production applications, and I’m passionate about delivering impact through data-driven solutions.

Available to hire

I’m a hands-on machine learning engineer who loves turning messy data into actionable insights. As a team lead at iCog Labs and SingularityNET, I enjoy shaping research directions and building scalable AI systems that perform in real-world settings. My work focuses on efficient model architectures, robust evaluation, and pragmatic tooling to reduce energy consumption and training time while preserving model quality and interpretability.

Earlier in my career, I built a full-stack web tool and gained experience across the software stack, from Laravel-based backends to admin dashboards. I thrive in collaborative environments, bridging research and production applications, and I’m passionate about delivering impact through data-driven solutions.

See more

Experience Level

Expert
Expert
Expert
Expert

Language

Amharic
Advanced
English
Advanced

Work Experience

Machine Learning Engineer (Team Lead) at iCog Labs, SingularityNET
July 1, 2024 - Present
Headed a 5-person research team developing predictive coding transformer architecture, reducing training energy from 14.0 to 1.74 (87% improvement) and perplexity from 50,000 to 309.6k through Hebbian learning algorithms. Architected PC-Transformer with lateral connections and energy-based inference, matching backpropagation accuracy while demonstrating biological plausibility in neural computation, evaluated on datasets ranging from 1M to 200M tokens. Benchmarked Neural Generative Coding models against traditional backprop on MNIST, demonstrating superior early-epoch performance and biological learning principles. Optimized the training pipeline implementing Distributed Data Parallel processing, memory management routines, and hyperparameter tuning, reducing training time by 40%. Researched MeTa pattern mining project implementing empirical truth-value functions and utility modules, translating C++ and Scheme implementations to a functional paradigm.
Website Developer (Intern) at Vintage Technologies
August 1, 2022 - October 1, 2022
Developed a full-stack task management platform using Laravel MVC framework and Bootstrap enabling task assignment, deadline tracking, and team collaboration. Implemented admin dashboard with role-based access control, project timeline visualization, and automated deadline notifications.

Education

Master of Artificial Intelligence and Machine Learning at Adelaide University
February 1, 2026 - March 9, 2026
Bachelor of Science in Computer Science at University of Gondar
October 1, 2019 - October 1, 2023

Qualifications

NeuroMath Academy Deep Learning Summer School
January 1, 2024 - March 9, 2026
European Summer School on Information Retrieval
January 1, 2024 - March 9, 2026
Oxford Machine Learning – Representation Learning and Generative AI specialization
January 1, 2024 - March 9, 2026
Deep Learning Specialization – CourseRa
January 1, 2024 - March 9, 2026
Machine Learning Specialization – CourseRa
January 1, 2024 - March 9, 2026

Industry Experience

Software & Internet, Professional Services, Media & Entertainment, Education