Projects

Artisan Bio: Director, AI and Computational Biology

At Artisan Bio, I am managing a team of expert computational biologists and AI researchers working on innovative solutions for genome engineering and target discovery for CRISPR based cell therapy and design. I am accountable for conception, design, and leadership of multiple complex projects spanning genomics, data science, cell analytics and synthetic biology while effectively leveraging subject matter experts in a cross-functional manner. I serve as a thought leader and providing technical advice on bioinformatics, computational biology, and AI at Artisan. As a leader, I am responsible for strategically planning and prioritizing team activities to further company goals, and effectively identifying risks and developing mitigation strategies for AI/computational biology.  I also develop software tools and contribute to bioinformatic and ML analyses, and oversee activities to deliver key results for Artisan’s genome editing platform and to our collaborators.

Invitae: Group Lead, Computational R&D

I led a team of bioinformaticians in using machine learning, statistics and novel algorithms to detect genomic biomarkers in cancer and other genetic diseases. I also served as the lead of technical leads for cross-functional collaborations to create and upgrade algorithms to detect genomic biomarkers for cancer applications. A few projects I contributed to include Tumor Mutational Burden (TMB) detection, AML-MRD pipelines and a novel CNV caller.

SomaLogic: Bioinformatics data scientist

I led the first international launch of SomaSignal tests in collaboration with NEC corporation in Japan. I developed multiple disease risk and lifestyle-based outcome models using AI and machine learning techniques with large-scale proteomic and clinical data. These models were critical components of our flagship product. I led a project to develop tools to assess model stability and efficacy. I also worked on deep learning techniques for proteomics data, and advancing survival analysis techniques by applying machine learning components. I have two published patents from SomaLogic:

Post Doc: Inferring relationships between genetic variants and smoking traits

Using GWAS techniques, I found correlations between genetic variants and smoking traits in large datasets with ~500,000 individuals. I developed solutions for machine learning-based fine-mapping to understand the genetic architecture of common loci identified by large scale GWAS.

Ph.D. Thesis: Predicting drug resistance in Mycobacterium Tuberculosis               

My research involved using machine learning and genomic analysis to predict drug resistance in Mycobacterium tuberculosis (M. tb). I created a fully automated genome sequence analysis and mutation annotation pipeline to characterize existing mutations in M. tb genomes. With a training set of ~3600 strains, I created a diverse feature set and novel feature selection and classification techniques for such datasets. Finally, to ensure that our predictive models reach the clinical-user population, I made a web application to facilitate fast delivery of drug-resistance profiles.

TellApart: Recommender systems in ad tech

I designed, implemented, tested and analyzed a purchase-to-purchase recommender system, which resulted in a 67% lift in conversions for transactional retargeting users. In addition, I worked on improving feature selection for the real-time production pipelines for scoring users and setting bids for retargeting ads.

My work at TellApart was done using Hadoop, Hive and MapReduce on big data.

Oncomine 

Adjustable autonomy for unmanned ground vehicles