Bereket B. Kinfu

I’ve always been the type of person who actually enjoys getting stuck on a difficult problem. For me, there’s a specific kind of satisfaction in deconstructing a complex mathematical proof or spending a weekend morning with a research paper that’s just outside my comfort zone.

I consider myself a math and research aficionado, but I’m most at home when I can take those theoretical concepts, things like the underlying calculus of a model or the mechanics of a new paper, and actually build something tangible with them. I don't just want to know that an algorithm works; I want to understand the 'why' behind it so I can make it more efficient or adapt it to a new challenge.

That curiosity is what drives my work in machine learning. I spend a lot of my time bridging the gap between what's happening in research and how we can apply it to solve real problems. I’m looking for a team where that kind of deep-dive thinking is encouraged and where I can help turn those complex ideas into something impactful

Machine Learning Engineer and Mathematical Researcher at iCOG-LABS (2025-03 – Present)

Machine Learning Engineer and Mathematical Researcher leveraging advanced AI techniques for reasoning engines and NLP models.

• Utilized Probabilistic Logic Networks (PLN) and Graph Theory to design reasoning engines within the OpenCog Hyperon framework.
• Investigated Information Theory and Markov Processes to refine Natural Language Understanding (NLU) models for low-resource languages (e.g., Amharic, Tigrinya).
• Implemented a Bayesian Optimization framework for hyperparameter tuning, outperforming standard grid search and saving 40+ hours of GPU compute time per month.
• Quantized models using TensorRT for deployment on edge devices, maintaining 98% accuracy while achieving sub-30ms latency.

B.S. in Electrical Engineering (Concentration: Computer Engineering) – Addis Ababa Science and Technology University