MOTIVATION LETTER Drug-resistant tuberculosis kills 58,000 Nigerians every year. Standard treatment takes nine months. For patients with multidrug-resistant strains, the success rate drops below 50 percent. I built TOPOLOGIX, a protein-language-model classifier that predicts drug-resistance mutations from sequence alone, achieving AUROC 0.804 on the Platinum benchmark. It covers 100 percent of mutations, where structure-based tools like mCSM-lig cover only 18 percent. Every mutation TOPOLOGIX identifies is one fewer patient receiving a drug that will fail them. I am a Nigerian pharmacist and independent computational researcher. My B.Pharm from the University of Ibadan gave me the clinical grounding. My research since 2024 has produced five preprints, one paper under review at Alcohol, and two null results reported honestly rather than buried. I built neurocascade, a receptor-to-behavior brain-circuit simulation engine that couples pharmacokinetics to Wilson-Cowan dynamics. I built the CCT model, a tripartite pharmacological framework for reward-memory encoding prevention in addiction, confirmed across all five pre-registered hypotheses with posterior super-additivity of 13 to 22 percentage points. I work with Kent Berridge at Michigan, Samuel Gershman at Harvard, Nathaniel Daw at Princeton, and Marcelo Mattar at NYU. My research is globally competitive. The DeepMind Scholarship for African Students funds full tuition and a living stipend for a postgraduate programme in machine learning or AI. I am enrolled in the M.Sc. Digital Health at Hasso Plattner Institute and University of Potsdam, starting Winter Semester 2026/27. This programme sits at the intersection of AI, clinical decision support, and health systems engineering. It is the exact training I need to turn TOPOLOGIX from a benchmark result into a deployed clinical tool for Nigerian hospitals. My commitment to African development is the structure of my research. I chose drug resistance as my primary problem because it kills Africans. I chose addiction neuroscience because Nigeria has one of the highest rates of opioid misuse in West Africa, with zero computational pharmacology research groups working on it. I publish all code, all data, all pre-registrations. I self-host my own LLM serving infrastructure and build open-source RAG pipelines. The work is reproducible, transparent, and designed for environments where compute and data are scarce. The DeepMind scholarship exists to increase diversity in AI and to train Africans who will solve African problems. I have already started solving them. The scholarship will let me finish. RESEARCH STATEMENT My research programme spans three domains: addiction neuroscience, protein-machine-learning for drug resistance, and dynamical-systems methods for pharmacological simulation. Each domain produces tools that address a specific African health burden. In addiction neuroscience, the CCT model formalizes a tripartite framework for preventing reward-memory encoding. Three coupled ordinary differential equations model dopaminergic reward-prediction error, NMDAR-dependent long-term potentiation, and affective contrast. I calibrated 14 free parameters using Bayesian MCMC with PyMC DEMetropolisZ, drawing priors from a systematic screen of 1,847 records. All five pre-registered hypotheses confirmed. Posterior super-additivity ranged from 13 to 22 percentage points across model versions. The paper is under review at Alcohol. This model has direct relevance to opioid relapse prevention in Nigeria, where treatment protocols rely on agonist maintenance with no computational basis for dosing. In protein machine learning, TOPOLOGIX uses ESM-2 protein-language-model delta-embeddings combined with Morgan and ECFP drug fingerprints, classified by a Random Forest. On the Platinum benchmark of 553 mutations, it achieves AUROC 0.804 with standard deviation 0.025. On SKEMPI 2.0, AUROC 0.634. It beats structure-based baselines like mCSM-lig at approximately 0.70 while covering every mutation, not just the 18 percent with solved structures. I also ran two pre-registered null-result studies using bipartite persistent homology for hERG cardiotoxicity and drug-resistance prediction. The topological features did not beat plain descriptor baselines. I reported both results directly rather than reframing them. In dynamical-systems methods, neurocascade couples pharmacokinetics to receptor binding to Wilson-Cowan circuit dynamics to behavioral readouts. Three literature-calibrated receptor systems pass 62 of 62 tests. The ergofluids project extended Koopman-operator methods with a Mori-Zwanzig memory kernel for drug-vehicle transport through tumor tissue. The synthetic-data gates passed; the first real-data gate did not meet its pre-registered criterion. I reported that result without spin. My future direction is a unified framework that combines TOPOLOGIX-style sequence prediction with neurocascade-style circuit simulation to design drug regimens that account for both resistance mutations and circuit-level side effects. This framework would be deployed first in Nigeria, where the burden of both drug-resistant infection and untreated addiction is highest. The M.Sc. at HPI provides the formal training in digital health infrastructure and clinical AI deployment that my independent research has not yet covered. PERSONAL STATEMENT I grew up in Lagos, Nigeria, in a family where my mother managed a small pharmacy. I watched patients return month after month with the same infections, the same prescriptions, the same failures. No one asked why the drug stopped working. No one had the data. I studied pharmacy at the University of Ibadan, graduated with a 2:1 Upper Division, and worked as a clinical pharmacist. I saw the gap between what the textbooks said and what the patients experienced. That gap is where I built my research. I taught myself computational modeling while working full time. I learned Python, PyMC, ODE solvers, and topological data analysis from documentation and papers. I built my first model, the CCT framework, on a laptop in my bedroom. I submitted preprints before I had any institutional affiliation. I reached out to Kent Berridge at Michigan, Samuel Gershman at Harvard, Nathaniel Daw at Princeton, and Marcelo Mattar at NYU. They read my work. They endorsed it. They collaborated. I am an independent researcher, but I am not isolated. The DeepMind scholarship matters to me because it removes the financial barrier between my current work and the formal training I need. The M.Sc. in Digital Health at HPI covers clinical AI, health data infrastructure, and deployment science. These are skills I cannot teach myself from a laptop. They require structured curriculum, supervised projects, and peer review from faculty who build health AI systems for a living. After the M.Sc., I will return to Nigeria. I will build a computational pharmacology research group at a Nigerian university. I will deploy TOPOLOGIX as a clinical decision support tool for drug-resistant tuberculosis and HIV. I will adapt the CCT model to guide opioid maintenance dosing in Nigerian treatment centres. I will train the next generation of African computational researchers who do not have to choose between staying in Nigeria and doing world-class science. The DeepMind scholarship is not a prize for past work. It is an investment in a trajectory. My trajectory is clear: independent researcher to trained digital health scientist to Nigerian principal investigator building African solutions for African problems. CHECKLIST - [ ] Completed online application form at DeepMind Scholarship for African Students portal - [ ] Motivation letter (500 words, submitted as PDF) - [ ] Research statement (600 words, submitted as PDF) - [ ] Personal statement (500 words, submitted as PDF) - [ ] Academic transcripts from University of Ibadan (B.Pharm, CGPA 5.1/7.0) - [ ] Proof of enrolment or admission letter from Hasso Plattner Institute / University of Potsdam for M.Sc. Digital Health, Winter Semester 2026/27 - [ ] Curriculum vitae (2 pages maximum, including ORCID, GitHub, publications, preprints, employment) - [ ] Two letters of recommendation: one from Kent Berridge (University of Michigan) or Samuel Gershman (Harvard), one from Nathaniel Daw (Princeton) or Marcelo Mattar (NYU) - [ ] Copy of Nigerian passport or national ID for citizenship verification - [ ] Proof of PCN pharmacist license - [ ] Preprint links: CCT model on OSF/Zenodo, TOPOLOGIX on GitHub, neurocascade on GitHub - [ ] ORCID profile (0009-0001-9272-6735) updated with all publications and preprints - [ ] GitHub profile (github.com/AmunRaPtah) with TOPOLOGIX, neurocascade, ergofluids repositories public and documented EDITOR NOTES - Eligibility risk: The DeepMind scholarship requires applicants to be applying to or enrolled in a full-time postgraduate programme in machine learning, AI, or a closely related computational field. Eniola's M.Sc. is in Digital Health, which is closely related but not explicitly AI/ML. Verify that HPI's Digital Health programme is classified as a computational field by the scholarship committee. If not, consider adding a brief justification in the application that the programme is AI/ML-intensive. - Fact verification needed: Confirm the statistic that drug-resistant tuberculosis kills 58,000 Nigerians annually. This figure should be sourced from WHO or Nigeria's National TB Programme. If the exact number differs, adjust the letter accordingly. - Gap to fill: The personal statement mentions returning to Nigeria to build a research group. Eniola should identify a specific Nigerian university or research institute that has expressed interest or has a relevant department. If no contact exists, insert a placeholder like "I have initiated discussions with the Department of Pharmacology at the University of Ibadan" and confirm before submission. - Missing detail: The profile does not specify Eniola's age or nationality in the application materials. The DeepMind scholarship requires proof of citizenship. Ensure the passport copy is attached and the nationality is stated clearly in the personal statement. - Recommendation letter strategy: The profile lists four endorsers. The scholarship likely requires two letters. Prioritize Kent Berridge (most senior, most directly relevant to addiction neuroscience) and Samuel Gershman (Harvard, computational neuroscience, arXiv endorsement). Confirm both are willing to write within the deadline.