DrugnomeAI – Gene Druggability Prediction
Published ML framework predicting druggability across all 19,846 human protein-coding genes; integrated 324 features from 15 sources and outperformed three published benchmarks.
Projects
Multivariate ICU Forecasting with StemGNN
The full clinical time-series pipeline behind my PhD: 4,120 raw admissions filtered to 364 high-quality patients and ~12.5M observations of 5-second bedside-monitor data.
Adversarial LLM evaluation for expert scientific reasoning
Adversarial tasks benchmarking frontier LLM agents on expert bioinformatics and scientific coding, with transcript audits before submission.
PubMed-to-Elasticsearch data pipeline
Indexes 35M+ biomedical citations and abstracts from the NLM PubMed baseline, with NLP pipelines that identify, prioritise, and validate candidate drug–target–disease relationships.