IMPROVE Project @ Frederick National Laboratory

Accomplishments:

  • Drove the final development of the IMPROVE project, a benchmarking framework for supervised learning on tabular data, with an initial focus on precision oncology applications
  • Designed and implemented major architectural changes to the improvelib Python package to enhance capabilities and accommodate varying applications
  • Authored user and developer documentation for improvelib and associated benchmarking workflows, enabling ease of adoption
  • Served as the domain expert in cancer biology, assessing model inputs and predictions for consistency with known biological mechanisms and therapeutic relevance
  • Assessed model generalization across heterogeneous drug response datasets, evaluating whether models learn meaningful biological signals, and explored the use of synthetic data to enhance feature utilization
  • Evaluated and curated single-agent and combination drug response AI/ML models into IMPROVE, ensuring reproducibility and compliance with benchmarking criteria
  • Developed modular, end-to-end workflows for model evaluation, enabling consistent comparison across datasets and model types
  • Executed and validated machine learning models using the CoderData benchmark dataset to assess performance and generalizability

You can view the documentation for the IMPROVE project here.

Publications from this work can be found here

Nussenzweig Lab @ National Cancer Institute

Accomplishments:

  • Identified targets for synthetic lethal interactions with transposon activation in cancer by mining multi-omic datasets
  • Identified mechanisms of resistance to WRN inhibition in cancers with microsatellite instability
  • Established protocols for execution and analysis of CRISPR/Cas9 knockout screens
  • Characterized H-DNA at homopurine/pyrimidine repeats in human cells, leading to a model of H-DNA formation during replication, by integrating S1-END-seq data with meta-analysis of Okazaki fragment and replication timing sequencing, as well as International Cancer Genome Consortium and COSMIC mutation data

Publications from this work can be found here.

Smith Lab @ New York University

Accomplishments:

  • Determined the role of limiting concentrations of lagging strand polymerases on Okazaki fragment synthesis
  • Created custom scripts and pipelines for integration of sequencing analysis with published datasets
  • Optimized a range of techniques, from CRISPR/Cas9 genome engineering to novel sequencing-based methods, as well as computational analysis
  • Taught a range of classes, from introductory biology to lab-based classes, with excellent evaluations

Publications from this work can be found here.

Baselga Lab @ Memorial Sloan Kettering Cancer Center and Massachusetts General Hospital

Accomplishments:

  • Examined inhibition of DNA repair, particularly homologous recombination, as a radiosensitization strategy in preclinical models of head and neck cancer
  • Investigated resistance to PI3K inhibitors, receptor tyrosine kinase inhibitors, and estrogen receptor inhibitors leading to high impact publications
  • Independently managed all aspects related to day-to-day functioning of a highly productive lab, including equipment, supplies, reagents, animals, and licenses, as well as mentoring and supervising multiple junior technicians.
  • Established the first patient-derived xenograft program at MSKCC for breast and head & neck cancer
  • Seamlessly transitioned the laboratory from Boston to New York in less than a week

Publications from this work can be found here.