- Conceptualized, trained, validated, and productionalized a novel non-tumor variant detection model for xM; Tempus’s liquid biopsy MRD test.- Led the development and CAP/CLIA validation of a chromosomal arm-level alteration detection ML algorithm based on DNA sequencing.- Spearheaded the platform migration from AWS to GCP and expanded the Tumor of Unknown Origin algorithm.- Delivered strategic analyses to pharmaceutical clients.- Delivered talks to external pharmaceutical strategic collaborators and internal cross-team audiences.- Collaborated with the commercial business unit to create machine learning based helper models.

Diagnostic, prognostic, and predictive multimodal ML model development for various oncology datasets. CAP/CLIA analytical validation for algorithmic biomarker tests. Real Word Data precise cohort identification. Risk/response/survival analysis.

Pioneering startup in precision psychiatry, developing new targeted pharmacological treatments for mental disorders.My role was designing and implementing machine learning solutions for detecting mental health genetic and neural biomarkers:- Led the development of Supervised Machine Learning models to utilize EEG and genetics features in Precision Psychiatry, predicting responses to various antidepressant treatments.- Directed the genomics team's codebase design and implementation, including data ingestion, genotype calling and QC, polygenic risk scoring, GWAS, ancestry estimation, and genotype imputation.- Conducted statistical inference, machine learning, feature extraction, and visualization for various psychiatric EEG and genetics datasets, providing weekly updates to leadership.

- Developed a software package for cerebellar Purkinje cells' spike sorting, achieving usable accuracy using unsupervised learning.- Developed a package for analysis of behavior driven cerebellar neural data (Python, Jupyter, Docker).- Designed and implemented protocols for data synchronization in a novel NHP electrophysiology recording system (C++, Open Ephys, Python).- Performed acute multichannel electrophysiological recording and behavioral training of animal models.- Developed a software package for neuronal axon tractography in 3D fluorescent images, obtained using light sheet microscopy on cleared brain tissues.

Systems Neural Engineering Lab: http://snel.gatech.edu/- Used novel deep learning methods for optimizing performance of various neural signal decoders.- Implemented a classifier to predict recall in memory given the stereoelectroencephalography recordings during memory encoding period (Using data from the Restoring Active Memory (RAM) project).- Implemented a regression model to predict hand movement based on microelectrode array recordings from motor cortex of non-human primates.- Set up the lab's network infrastructure: central authentication and user management, central file share system, firewall management, etc. all in CentOS workstations.- Trained rat models for performing supination and pronation tasks.

The Miller Lab: http://millerchembio.com/• Developed spikeNet, a neuronal network analysis GUI-enabled software for voltage imaging experiments, written in Matlab. Incorporated unsupervised machine learning algorithms and novel data correlation analysis techniques to allow for high throughput study of voltage imaging data by minimizing user inputs. Performed a developmental study on rat embryonic hippocampal neurons (in vitro) using this tool. Presented parts of this program at the BIC Symposium 2016, UCSF. Published in the journal Frontiers in Neuroscience.• Developed spikeMapper, a voltage imaging spike detection and spike sorting program written in Matlab. Used regression and statistical analysis methods to streamline spike detection and producing relevant raster plots, heat maps, trace plots, etc. It was used in a published study in PNAS.• Developed LiKa, an automated cell detection and fluorescent emission analysis tool used for characterization of newly synthesized voltage sensitive dyes. Presented in the College of Chemistry annual research symposium and was acknowledged in the corresponding paper in JACS.• Wrote a camera trigger and LED timing control software using Micro Manager’s Java API for a custom built fluorescent microscope system.• Performed various lab tasks: Imaging, Cell passage, Solution preparation, etc.

Molecular Cell Biomechanics Lab (Mofrad Lab): http://biomechanics.berkeley.edu/• Worked on development of a benchmark for protein protein interaction prediction methods.• Wrote Python scripts to automatize protein interaction data collection and extraction of desired features• Wrote Python scripts for automatic protein network extraction from collected data

• Conducted more than 1800 EME (Electromagnetic Emission) surveys by performing near field measurements at cell sites in more than 30 states across the country. Generated FCC regulated EME surveys using the measured values as well as simulated situations.