MOTIVATION LETTER Open Philanthropy funds work that identifies and solves neglected problems at scale. Antimicrobial resistance kills an estimated 1.27 million people per year, and the World Health Organization lists it among the top ten global health threats. Sub-Saharan Africa carries a disproportionate share of that burden, yet most resistance-prediction tools fail on African clinical contexts because they depend on protein structures that exist for only about 18 percent of clinically relevant mutations. My research addresses exactly this gap. I am Eniola Ayodele Olutogun, a pharmacist and independent computational researcher based in Nigeria, enrolled in the M.Sc. Digital Health program at the Hasso Plattner Institute and University of Potsdam. My current project, TOPOLOGIX, predicts drug-resistance mutations from protein sequence alone using ESM-2 protein language model delta-embeddings combined with Morgan fingerprints and a Random Forest classifier. On the Platinum benchmark of 553 mutations, TOPOLOGIX achieves an AUROC of 0.804 plus or minus 0.025, outperforming the structure-based baseline mCSM-lig at approximately 0.70 while covering 100 percent of mutations. This means the tool works where structure-based methods cannot. Open Philanthropy's selection criteria emphasize quantitative rigor, independent thinking, and demonstrated ability. My research record reflects all three. I pre-registered and powered a replication study testing whether bipartite persistent homology predicts hERG cardiotoxicity; the result was negative, with topological features scoring 0.8426 AUROC against 0.8782 for a plain descriptor baseline. I reported the null result directly rather than reframing it. That study settled a comparison the published literature had never actually run. I applied the same topological methods to drug-resistance prediction, found they carried almost no signal, and pivoted to the sequence-representation approach that became TOPOLOGIX. My CCT model, a tripartite pharmacological framework for reward-memory encoding prevention in addiction, confirmed all five pre-registered hypotheses through Bayesian MCMC calibration with 14 free parameters and literature-elicited priors from a screen of 1,847 records. Open Philanthropy values work that can inform policy and drug development. TOPOLOGIX is built as an end-to-end pipeline, from raw sequence data to interpretable predictions, using open-source tools and reproducible methods. It is designed to be deployed in low-resource settings where sequencing data exists but structural biology does not. This aligns directly with Open Philanthropy's focus on global health and its commitment to interventions that work in the real world. I work remotely and independently by necessity. I have built four DuckDB-based ingest-to-analyze pipelines across life sciences, technology, and social science domains, self-host local LLM serving infrastructure, and manage production systems including Linux VPS, CI/CD, and automated backup. I am accustomed to the discipline of asynchronous collaboration and clear written communication. Open Philanthropy's remote-first model and its willingness to fund independent researchers make this an exceptional fit. I am applying for the x3 Global Remote Job Positions because my research trajectory, my demonstrated record of falsifiable pre-registered science, and my technical capacity to build and deploy tools from sequence to prediction match what Open Philanthropy needs. RESEARCH STATEMENT My research program addresses a specific problem: predicting drug-resistance mutations from biological sequence data in settings where structural information is unavailable. This problem sits at the intersection of global health, machine learning, and computational biology, and it is the direct focus of my current project, TOPOLOGIX. The motivation is epidemiological. Antimicrobial resistance is a leading cause of death globally, and the burden falls hardest on low- and middle-income countries. Resistance arises from mutations in drug targets, and predicting which mutations will confer resistance enables earlier intervention, better drug design, and more informed policy. The dominant tools for this task, such as mCSM-lig, require a protein structure. But structures are unavailable for most clinically relevant mutations. In the Platinum benchmark, structure-based tools cover roughly 18 percent of mutations. The remaining 82 percent are invisible to them. TOPOLOGIX removes that dependency. The method uses ESM-2 protein language model delta-embeddings to represent the sequence context of each mutation, combines these with Morgan/ECFP drug fingerprints, and trains a Random Forest classifier to predict resistance. On the Platinum benchmark of 553 mutations, TOPOLOGIX achieves an AUROC of 0.804 plus or minus 0.025, beating mCSM-lig at approximately 0.70 while covering 100 percent of mutations. On SKEMPI 2.0, the AUROC is 0.634. The performance gap between structure-based and sequence-based methods closes, and the coverage gap disappears entirely. The research path to TOPOLOGIX was deliberately falsifiable. I first tested whether interface topology, measured through bipartite persistent homology with an opposition-distance metric, could predict hERG cardiotoxicity from protein-ligand interface geometry. The pre-registered, powered replication found that topological features did not beat a plain descriptor baseline, 0.8426 versus 0.8782 AUROC. I then applied the same topological constructs to drug-resistance prediction on the Platinum benchmark. The result was negative: AUROC of 0.425 and 0.485, ruling out interface geometry as the driver of resistance signal. These two negative results motivated the sequence-representation approach that became TOPOLOGIX. I report the negative results as published preprints because they settle questions the literature had never actually tested. My broader research program applies the same standard of pre-registration and honest reporting. The CCT model, a tripartite pharmacological framework for reward-memory encoding prevention in addiction, couples dopaminergic reward prediction error, NMDAR-dependent long-term potentiation, and affective contrast in an ODE model calibrated with Bayesian MCMC. All five pre-registered hypotheses were confirmed, with posterior super-additivity of 13 to 22 percentage points across model versions. The neurocascade project builds a receptor-to-behavior brain-circuit simulation engine coupling pharmacokinetics to receptor binding to Wilson-Cowan circuit dynamics, with 62 of 62 tests passing. The ergofluids project extends Koopman operator methods with a Mori-Zwanzig memory kernel to model drug transport through tumor tissue; its first real-data validation gate did not meet its pre-registered criterion, and I reported that result directly. For Open Philanthropy, the relevant output is TOPOLOGIX. It is a tool that works where existing methods fail, it is built on open-source components, and it is designed for deployment in low-resource settings. The next steps are validation on African clinical resistance datasets, extension to additional drug classes, and integration into surveillance pipelines. The work is independent, pre-registered, and reproducible, which matches Open Philanthropy's emphasis on rigorous, high-impact research. EDITOR NOTES - Research line selected: TOPOLOGIX. This is the only active project that directly matches Open Philanthropy's stated interests in AI-driven scientific discovery and global health, specifically antimicrobial resistance. The hERG topology study and the resistance-topology study are cited as negative results that motivated TOPOLOGIX, not as current work. The CCT, neurocascade, ergofluids, and psyche-twin projects are mentioned only as evidence of methodological range and honest reporting, not as the primary pitch. - Eligibility risk: The posting is for "x3 Global Remote Job Positions," which may be employment rather than a fellowship or grant. The applicant is based in Nigeria and enrolled in a German M.Sc. program. Confirm whether Open Philanthropy can employ someone in Nigeria, or whether the role is structured as a contractor or fellowship arrangement. If employment is not possible, the application should clarify the applicant's willingness to work as an independent contractor. - Fact verification needed: Confirm the exact wording of the job posting, including whether there are multiple distinct roles or one role with three openings. Confirm the AUROC figures for TOPOLOGIX (0.804 plus or minus 0.025 on Platinum, 0.634 on SKEMPI 2.0) and the mCSM-lig baseline of approximately 0.70. Confirm the coverage claim of 100 percent versus 18 percent for structure-based tools. Confirm the hERG AUROC figures (0.8426 versus 0.8782) and the Platinum topology AUROC figures (0.425 and 0.485). - Gap to fill: The applicant must insert a paragraph on why Open Philanthropy specifically, beyond the general fit, and what the applicant hopes to contribute to the organization's internal research or operations. The current draft focuses on what Open Philanthropy can do for the applicant's research; the letter should also address what the applicant can do for Open Philanthropy's mission, such as contributing to their global health or AI strategy work. - Gap to fill: The applicant must confirm availability for a remote role, time zone overlap with the Open Philanthropy team, and any visa or work authorization constraints. The letter should also state whether the M.Sc. program at HPI/Potsdam permits concurrent employment. - Gap to fill: The applicant must provide at least one named reference from the collaborators listed (Berridge, Gershman, Daw, Mattar) who can speak to the TOPOLOGIX work specifically, not just to the CCT model or general research ability.