I am a senior computational chemist, AI researcher and scientific software engineer. My research focuses on developing data-driven and physics informed models for porous materials with particular emphasis on metal-organic frameworks, covalent organic frameworks, batteries and energy-storage systems.
I combine first-principle DFT simulations with machine learning and scientific software development to accelerate materials discovery, crystal structure prediction, the design of models for predicting synthesis conditions, the elucidation of materials properties and the explanation of unprecedented synthetic outcomes.
My current ambition is to establish a start-up focused on building end-to-end digital chemistry workflows for the automated discovery, synthesis and scale-up of advanced materials. The goal is to integrate data-driven modelling, artificial intelligence, robotics and scientific software to accelerate materials development for applications in batteries, hydrogen storage and CO2 capture.
In parallel, I am continuing my academic research as a postdoctoral researcher, where I study the effect of functional group isomerism on gas adsorption and diffusion in metal-organic frameworks.
I was born and raised in Cameroon, where I completed my primary and secondary education and later earned both a B.Sc. and M.Sc. in Chemistry. These years shaped my love for science and fueled my desire to further understand how and why things happen the way they do.
I moved to New Zealand for my PhD in Chemistry at the University of Canterbury, where I built a quantum-mechanical model to predict molecules susceptible to undergo light-induced cyclisation reactions.
You can can download my PhD thesis from the following link PhD Thesis
One thing that I learned from my PhD was that that quantum-mechanical models alone are insufficient to overcome the conventional trial-and-error processes involved in laboratory synthesis.Thus to truely accelerate and scale bench synthesis, it is essential to integrate data-driven techniques, artificial intelligence and automation with quantum mechanical modelling.
For this reason, I pursued a postdoctoral position at Nottingham Trent University in the UK. During this period, I designed new stationary phases for the chromatographic separation of peptides and fullerenes, performed crystal structure prediction of covalent organic frameworks by modeling their stacking arrangements, and became actively involved in data science through the processing and analysis of datasets comprising more than 80 metal–organic frameworks. Despite the challenges posed by the COVID-19 pandemic, I successfully published eight peer-reviewed articles, taught scientific Python to first-year chemistry students, and supervised several undergraduate and postgraduate research projects.
I later moved to the Karlsruhe Institute of Technology (KIT) in Germany to pursue a second postdoctoral position. During this postdoc, I led the FAIRMat project in Area E5 as a software developer, where I curated and built a FAIR-compliant database of over 40,000 metal–organic frameworks. This work involved meticulous correction of crystal structures, including the addition of missing hydrogen atoms, removal of unbound guest species, filtering of structures with atomic overlaps, and subsequent geometry optimisation of the cleaned systems.
In parallel, I significantly advanced my scientific software development skills by implementing a production-grade Python module for the automated deconstruction of MOFs into unique building units. The module enables extraction of cheminformatic identifiers, computation of porosity, identification of open metal sites, and classification of secondary building unit topologies. This software formed the foundation of the Porosity Normaliser in NOMAD, used to classify and process uploaded structures as MOFs, COFs, or zeolites.
During this period, I also wrote my MSCA postdoctoral fellowship proposal, which was subsequently awarded with an evaluation score of 98%, reflecting both the scientific ambition and technical robustness of the proposed research.
During my Marie Skłodowska-Curie postdoctoral fellowship, I focused on bridging the gap between computational materials discovery and experimental synthesis. I extensively text-mined synthesis conditions from the scientific literature using natural language processing and transformer-based models, where I created a one-to-one mapping between curated crystal structures of metal–organic frameworks and their experimentally reported synthesis conditions.
This rich structured dataset enabled me to train graph neural network models for predicting synthesis conditions of MOFs directly from their three-dimensional crystal structures. This approach provides a critical step toward synthesis planning by significantly reducing the time spent on manual literature searches and enabling the experimental realisation of computationally discovered materials.
The motivation behind this work stems from the striking contrast between the millions of high-performing hypothetical MOFs reported for applications such as energy storage and separations and their very limited adoption at the industrial scale. This gap is largely driven by the high activation barriers associated with their synthesis, scalability and reproducibility. I view the next phase of materials discovery as the development of robust data-driven strategies that can accelerate synthesis and industrialisation to enable promising materials to move beyond computational predictions into real-world applications.
To deepen my understanding of practical synthesis constraints, I undertook a one-month research exchange in Karlsruhe, where I gained hands-on experience synthesising potassium-based cyclodextrin SURMOFs and developed a robotic workflow for accelerating and automating synthesis processes.
Alongside my research, I have led multiple interdisciplinary collaborations with experts in synthetic chemistry and artificial intelligence to accelerate the discovery and synthesis of advanced materials for applications including energy storage, pharmaceutical detection and extraction from water and isomer separations. I also supervise research projects, secure large-scale HPC allocations, presented my work at international conferences, and publish severapeer-reviewed articles and production grade software.
I am actively engaged in scientific software development, building Python modules, web interfaces, and desktop applications that support synthesis planning, database exploration, and reproducible materials research. I have presented this work at high-profile forums, including the Falling Walls Science Summit and was selected to participate in the Lindau Nobel Laureate Meetings.