• Contribute to AI safety research at a leading alignment-focused organization, studying how capabilities and values emerge during model training, with the goal of building scalable interventions to align frontier models
• Develop core research infrastructure and translate exploratory prototypes into robust, scalable systems

• Led a 5-member team and directed 12 research projects (4 independently, 8 collaboratively) at the intersection of AI safety and biology, developing novel methods to evaluate and restrict dual-use capabilities of agentic LLMs and biological foundation models — misuse potential, adversarial robustness, and synthetic biology risk
• Spearheaded red-teaming of agentic AI scientists, discovering novel vulnerabilities
• Evaluated adversarial robustness of genomic foundation models, including susceptibility to data poisoning
• Directed benchmarking of genomic language models against real genomes, identifying systematic discrepancies that can be used to detect synthetically generated sequences
• Developed tamper-resistant weight-locking methods for open-weight biological AI models

• Built a scalable High-Performance Computing pipeline in Slurm enabling the identification of pathogen biomarkers across 45,000 genome assemblies; led scaling to 500,000 assemblies
• Developed the kmerDB database consolidating short DNA and protein sequences across 60K+ genomes and proteomes; co-developed 4 additional public scientific databases
• Created the open-source ZSeeker tool; co-developed 5 additional open-source bioinformatics tools
• Supervised 8+ researchers and mentored 5 junior researchers to their first first-author publications
• Drafted grant applications resulting in $3M+ in competitive funding

• Co-founded a diagnostics startup, coordinating across technical, regulatory, and investor stakeholders to translate patented nullomer research into a clinical cancer-detection platform
• Developed an ML pipeline for cancer detection from liquid biopsies, achieving AUC ranging from 0.89 to 0.94
• Contributed to securing $850K through technology development and investor demos

• Optimized a cancer-genomics ML pipeline using Julia, achieving a 50-fold speedup and saving >3 months of computational time and $50K+ in costs
• Deployed and managed a genomic analysis pipeline on Google Cloud Platform, processing 14 TB of cfDNA sequencing data

Doctoral thesis: Computational and machine learning methods for k-mer sequence analysis in large-scale omics data

GPA: 4.0/4.0

Honours: Cum Laude

Master’s thesis: Frequentmers – a k-mer based approach for the detection of liver cirrhosis gut microbiome mNGS