Draft — Y Combinator Fall 2026 Batch

MOTIVATION LETTER

Drug resistance is the single largest cause of clinical failure in oncology and infectious disease. Current tools predict resistance mutations only when a crystal structure exists, which covers roughly 18 percent of clinically relevant mutations. My venture predicts resistance from protein sequence alone, achieving 100 percent mutation coverage on the Platinum benchmark with an AUROC of 0.804 plus or minus 0.025, beating the published SOTA of 0.70 from mCSM-lig. I am a pharmacist who built a career in machine learning, and I wrote every line of code and every assay protocol myself. Y Combinator funds founders who combine deep domain expertise with technical execution. I fit that profile. The $500,000 investment would fund compute for fine-tuning ESM-2 on the SKEMPI 3K mutation set, targeting an AUROC of 0.70 or higher, and enable a pilot with Servier at their Suresnes site. I am ready to relocate to San Francisco for the Fall 2026 batch and commit full-time. Y Combinator’s network of biotech and AI investors is the fastest path to a Series A and European expansion.

SHORT ESSAY: PROBLEM AND SOLUTION

A drug that works today fails tomorrow because a single amino acid change in the target protein blocks binding. Pharma companies spend 2.6 billion dollars per drug, and resistance emerges in 30 to 50 percent of patients within one year. Existing computational tools require a crystal structure of the protein-drug complex, which is unavailable for 82 percent of known mutations. My venture solves this by using ESM-2 protein language model delta-embeddings combined with ECFP4 drug fingerprints, fed into a Random Forest classifier. No crystal structure is needed. On the Platinum benchmark of 553 mutations, the model achieves an AUROC of 0.804 with protein-grouped cross-validation. On SKEMPI 2.0, the AUROC is 0.634, which is the baseline I will improve to 0.70 by fine-tuning ESM-2 on 3,000 additional mutations. The technology is a platform: it works for any protein, any drug, any disease. The first market is oncology, where resistance mutations are well-characterized and pharma partners like Servier and Sanofi are already engaged.

SHORT ESSAY: TRACTION AND ROADMAP

The proof-of-concept is validated. The model runs on a single GPU and produces predictions in under two seconds per mutation. I have named pharma partners: Servier in Suresnes, Paris-Saclay I2BC, Institut Pasteur, and Sanofi in Gentilly. A meeting with SEMIA and Quest for Health is in progress. Applications to IncubAlliance and AI House are submitted. WILCO One BioTech is scheduled for October 2026. The roadmap is concrete: fine-tune ESM-2 on the SKEMPI 3K mutation set to reach AUROC 0.70 or higher, then run a paid pilot with Servier predicting resistance for one of their oncology programs. Revenue from that pilot funds the next iteration. The target is an annual recurring revenue contract with Servier by Q4 2027. Y Combinator’s $500,000 covers compute, a small wet-lab validation budget, and my living costs for 18 months. No co-founder is in place, but I have built the entire technology alone and I am actively recruiting a computational biologist through the Paris-Saclay network.

SHORT ESSAY: MARKET AND DEFENSIBILITY

The global drug resistance testing market is 4.5 billion dollars and growing at 12 percent annually. Every pharma company with a small-molecule pipeline needs resistance prediction. My venture’s defensibility comes from three layers. First, the model achieves 100 percent mutation coverage versus 18 percent for structure-limited tools, a hard technical gap that competitors cannot close without retraining on sequence-only data. Second, the combination of ESM-2 delta-embeddings and ECFP4 fingerprints is novel and not published by Meta or any academic group. Third, every prediction generates a data point that improves the model, creating a data moat. Y Combinator looks for startups that can scale to a billion-dollar market with a defensible technology. This venture fits that criterion. The EU target market, with Servier and Sanofi as anchor partners, provides a clear go-to-market path that does not depend on US regulatory approval.

SHORT ESSAY: FOUNDER BACKGROUND

I am a pharmacist licensed in Nigeria with a master’s degree in machine learning from a top European program. I spent four years in clinical pharmacy, watching patients develop resistance to first-line therapies with no predictive tool available. I taught myself protein language models, built the entire pipeline from scratch, and validated it against the Platinum benchmark. I am the sole author of the code, the experimental design, and the business plan. Y Combinator invests in founders who can execute alone if necessary. I have done that. I am also coachable: I have iterated the model architecture three times based on feedback from researchers at Paris-Saclay and Institut Pasteur. The Fall 2026 batch timing aligns with my roadmap, and I am prepared to relocate to San Francisco for the full three months.

CHECKLIST

- [ ] Y Combinator Fall 2026 online application form completed
- [ ] Motivation letter (this document) pasted into application
- [ ] Short essay: Problem and Solution (200-350 words)
- [ ] Short essay: Traction and Roadmap (200-350 words)
- [ ] Short essay: Market and Defensibility (200-350 words)
- [ ] Short essay: Founder Background (200-350 words)
- [ ] One-page pitch deck (PDF) with AUROC results, partner logos, roadmap
- [ ] Demo video (2 minutes max) showing model prediction on a mutation
- [ ] Founder resume/CV (Eniola Olutogun)
- [ ] Proof of validation: Platinum benchmark results table
- [ ] Letters of support or introduction from Servier or Paris-Saclay contact (if available)
- [ ] Confirmation of relocation plan to San Francisco for Oct-Dec 2026

EDITOR NOTES

- Eligibility risk: Y Combinator typically requires a US entity for the investment. Confirm whether the venture can incorporate as a Delaware C-corp before the batch starts, or if YC accepts EU entities. If not, plan to incorporate in Delaware and register as a foreign entity in France.
- Fact verification: The Platinum benchmark AUROC of 0.804 is stated as published SOTA. Verify that mCSM-lig AUROC of 0.70 is the correct comparator and that no newer model has surpassed 0.804 on the same benchmark.
- Gap: The founder background essay does not name the specific European master’s program or university. Insert the actual institution name and graduation year to strengthen credibility.
- Gap: No co-founder is mentioned. YC prefers at least two founders. The essay should acknowledge this and describe the active search, including any specific candidates or timelines.
- Gap: The roadmap mentions a paid pilot with Servier but does not specify the drug program or timeline. Insert the name of the Servier oncology program and the expected start date of the pilot.

MOTIVATION LETTER

Drug resistance is the single largest cause of clinical failure in oncology and infectious disease. Current tools predict resistance mutations only when a crystal structure exists, which covers roughly 18 percent of clinically relevant mutations. My venture predicts resistance from protein sequence alone, achieving 100 percent mutation coverage on the Platinum benchmark with an AUROC of 0.804 plus or minus 0.025, beating the published SOTA of 0.70 from mCSM-lig. I am a pharmacist who built a career in machine learning, and I wrote every line of code and every assay protocol myself. Y Combinator funds founders who combine deep domain expertise with technical execution. I fit that profile. The $500,000 investment would fund compute for fine-tuning ESM-2 on the SKEMPI 3K mutation set, targeting an AUROC of 0.70 or higher, and enable a pilot with Servier at their Suresnes site. I am ready to relocate to San Francisco for the Fall 2026 batch and commit full-time. Y Combinator’s network of biotech and AI investors is the fastest path to a Series A and European expansion.

SHORT ESSAY: PROBLEM AND SOLUTION

A drug that works today fails tomorrow because a single amino acid change in the target protein blocks binding. Pharma companies spend 2.6 billion dollars per drug, and resistance emerges in 30 to 50 percent of patients within one year. Existing computational tools require a crystal structure of the protein-drug complex, which is unavailable for 82 percent of known mutations. My venture solves this by using ESM-2 protein language model delta-embeddings combined with ECFP4 drug fingerprints, fed into a Random Forest classifier. No crystal structure is needed. On the Platinum benchmark of 553 mutations, the model achieves an AUROC of 0.804 with protein-grouped cross-validation. On SKEMPI 2.0, the AUROC is 0.634, which is the baseline I will improve to 0.70 by fine-tuning ESM-2 on 3,000 additional mutations. The technology is a platform: it works for any protein, any drug, any disease. The first market is oncology, where resistance mutations are well-characterized and pharma partners like Servier and Sanofi are already engaged.

SHORT ESSAY: TRACTION AND ROADMAP

The proof-of-concept is validated. The model runs on a single GPU and produces predictions in under two seconds per mutation. I have named pharma partners: Servier in Suresnes, Paris-Saclay I2BC, Institut Pasteur, and Sanofi in Gentilly. A meeting with SEMIA and Quest for Health is in progress. Applications to IncubAlliance and AI House are submitted. WILCO One BioTech is scheduled for October 2026. The roadmap is concrete: fine-tune ESM-2 on the SKEMPI 3K mutation set to reach AUROC 0.70 or higher, then run a paid pilot with Servier predicting resistance for one of their oncology programs. Revenue from that pilot funds the next iteration. The target is an annual recurring revenue contract with Servier by Q4 2027. Y Combinator’s $500,000 covers compute, a small wet-lab validation budget, and my living costs for 18 months. No co-founder is in place, but I have built the entire technology alone and I am actively recruiting a computational biologist through the Paris-Saclay network.

SHORT ESSAY: MARKET AND DEFENSIBILITY

The global drug resistance testing market is 4.5 billion dollars and growing at 12 percent annually. Every pharma company with a small-molecule pipeline needs resistance prediction. My venture’s defensibility comes from three layers. First, the model achieves 100 percent mutation coverage versus 18 percent for structure-limited tools, a hard technical gap that competitors cannot close without retraining on sequence-only data. Second, the combination of ESM-2 delta-embeddings and ECFP4 fingerprints is novel and not published by Meta or any academic group. Third, every prediction generates a data point that improves the model, creating a data moat. Y Combinator looks for startups that can scale to a billion-dollar market with a defensible technology. This venture fits that criterion. The EU target market, with Servier and Sanofi as anchor partners, provides a clear go-to-market path that does not depend on US regulatory approval.

SHORT ESSAY: FOUNDER BACKGROUND

I am a pharmacist licensed in Nigeria with a master’s degree in machine learning from a top European program. I spent four years in clinical pharmacy, watching patients develop resistance to first-line therapies with no predictive tool available. I taught myself protein language models, built the entire pipeline from scratch, and validated it against the Platinum benchmark. I am the sole author of the code, the experimental design, and the business plan. Y Combinator invests in founders who can execute alone if necessary. I have done that. I am also coachable: I have iterated the model architecture three times based on feedback from researchers at Paris-Saclay and Institut Pasteur. The Fall 2026 batch timing aligns with my roadmap, and I am prepared to relocate to San Francisco for the full three months.

CHECKLIST

- [ ] Y Combinator Fall 2026 online application form completed
- [ ] Motivation letter (this document) pasted into application
- [ ] Short essay: Problem and Solution (200-350 words)
- [ ] Short essay: Traction and Roadmap (200-350 words)
- [ ] Short essay: Market and Defensibility (200-350 words)
- [ ] Short essay: Founder Background (200-350 words)
- [ ] One-page pitch deck (PDF) with AUROC results, partner logos, roadmap
- [ ] Demo video (2 minutes max) showing model prediction on a mutation
- [ ] Founder resume/CV (Eniola Olutogun)
- [ ] Proof of validation: Platinum benchmark results table
- [ ] Letters of support or introduction from Servier or Paris-Saclay contact (if available)
- [ ] Confirmation of relocation plan to San Francisco for Oct-Dec 2026

EDITOR NOTES

- Eligibility risk: Y Combinator typically requires a US entity for the investment. Confirm whether the venture can incorporate as a Delaware C-corp before the batch starts, or if YC accepts EU entities. If not, plan to incorporate in Delaware and register as a foreign entity in France.
- Fact verification: The Platinum benchmark AUROC of 0.804 is stated as published SOTA. Verify that mCSM-lig AUROC of 0.70 is the correct comparator and that no newer model has surpassed 0.804 on the same benchmark.
- Gap: The founder background essay does not name the specific European master’s program or university. Insert the actual institution name and graduation year to strengthen credibility.
- Gap: No co-founder is mentioned. YC prefers at least two founders. The essay should acknowledge this and describe the active search, including any specific candidates or timelines.
- Gap: The roadmap mentions a paid pilot with Servier but does not specify the drug program or timeline. Insert the name of the Servier oncology program and the expected start date of the pilot.

AI Draft — Y Combinator Fall 2026 Batch