The understanding and effective manipulation of biological systems, i.e. the ability to predict systems normal functioning and intervention outcomes, is a catalyser for advancing biomedicine and biotechnology, ultimately enabling the design of highly efficient therapies and engineering of microbial processes that can shift us from a fossil to sustainable society. Towards this, we couple computer and biological sciences by bridging state-of-the-art machine learning, bioinformatics, mathematical modelling together with experimental techniques to advance synthetic biology research. Our focus falls under the umbrella of two main directions: i) quantitative study of molecular dogma at the systems level and ii) AI-driven synthetic biology of DNA and proteins. In (i), we are interested in fundamental understanding the quantitative information that evolution encoded in biological sequences and learning from it for practical applications (ii) for designing DNA for expression systems in novel hosts, improving protein expression, protein fitness, and stability solubility optimisation for synthetic biology applications, e.g. biological degradation of plastics. We build mechanistic and data-driven machine learning models to learn quantitative genotype-phenotype relationships to understand molecular phenomena’ underlying mechanisms in natural systems. We then apply the gained knowledge and test predictions in experimental settings (mainly yeast and bacteria), e.g. by developing new synthetic molecules DNA/proteins based on machine learning designs for biotech applications.