Akbar Yermekov
CEO, Chief Data Scientist
Education
University of Oxford (UK)
Undergraduate Advanced Diploma in IT Systems Analysis and Design
Part-time studies, with distinction
MIT Institute for Data, Systems and Society (IDSS) & MITx
MicroMasters in Statistics and Data Science
University of the People (USA)
Associate of Science in Computer Science
Eastern Mediterranean University (Turkey)
Molecular Biology and Genetics (1st year, "High Honor")
Professional Experience
- •Full-stack data scientist with 12 years of experience
- •Expertise in machine learning, predictive modeling, generative and agentic AI, as well as software architecture and database design
- •Extensive experience designing & building predictive models, NLP and CV pipelines for various startups and international institutions
- •Developed and tested predictive models that were used by several Fortune 100 companies
Certificates
Utrecht University & Bijvoet Centre for Biomolecular Research
Exploring Nature's Molecular Machines
Summer School Program, 2014
Focus: Molecular Biology, Research Methods, Biological Data Analysis
University of Melbourne
Epigenetic Control of Gene Expression
Coursera Certificate with Distinction, 2014
Focus: Research Skills, Literature Reviews, Scientific Writing
University of Toronto
Bioinformatic Methods II
Coursera Certificate with Distinction, 2014
Focus: Structural Bioinformatics, Computational Genomics, R Programming, Sequence Alignment
Johns Hopkins University
Mathematical Biostatistics Boot Camp 1
Coursera Certificate, 2014
Focus: Statistical Modeling, Hypothesis Testing, Clinical Data Analysis
Duke University
Bioelectricity: A Quantitative Approach
Coursera Certificate, 2013
Focus: Research Skills, Biological Data Analysis
University of California, Berkeley
Quantum Mechanics and Quantum Computation
edX Certificate, 2013
Focus: Statistics, Applied Mathematics, Quantum Computing
Publications & Open-Source
- LibraryLibraryTITANS Trainer
GitHub · open source, 2026
Huggingface-style trainer for the TITANS architecture — transformers with neural long-term memory and test-time learning. Train TITANS models in ~5 lines of code.
- ModelModelBioTitan
Hugging Face · open weights, 2026
A foundational genomics model built on the TITANS architecture. Designed to learn long-range dependencies across DNA sequences and adapt at inference time via neural long-term memory.
- Paper
Yermekov A., Herrera-Martí D. A.
BIO Web of Conferences (eISSN: 2117-4458, Scopus) · peer-reviewed · in press, 2026
Accepted at the 16th International Conference on Bioscience, Biochemistry and Bioinformatics (ICBBB 2026), Kobe University, Japan — January 2026. Full paper & oral presentation.
- ArticleFrom the Age of Scaling to the Age of Research: How the Novel TITANS Approaches Shape BioTitan
LinkedIn, 2026
On why scaling alone is hitting its ceiling, how test-time learning and neural long-term memory change the picture, and how these ideas shape BioTitan — our open foundational model for genomics.
LinkedIn, 2024
An exploration of data leakage issues in AI model training and their implications.
Molkenov A., Zhelambayeva A., Yermekov A., et al.
Encyclopedia of Bioinformatics and Computational Biology · Elsevier, 2019
Co-authored chapter in the Encyclopedia of Bioinformatics and Computational Biology (Elsevier, 2019; eds. Ranganathan, Nakai & Schönbach), part of Elsevier's Reference Module in Life Sciences. Reviews transcriptomic databases across cancer research, human disease pathology, developmental biology, and broader repositories for animal and plant studies.
