San Francisco, California, United States
I have a master's degree in computer science from the University of Southern California, where I focused on data science and machine learning. I have been a software engineer for over four years, working in various domains such as healthcare, neuroscience, and finance. My core competencies include data pipelines, relational databases, and Python programming. Most recently, I was the lead engineer on a newly formed business payments team at Sana, a health insurance startup. I led the technical and product development and improvements, working closely with the product manager and other stakeholders. I value collaboration, feedback, and learning, and I enjoy working with diverse and talented teams. I am eager to apply my skills and experience to new challenges and opportunities in the software industry.
Lead technical and product development and improvements as lead engineer on a newly made business payments team. Prioritized and refined our backlog with the product manager. Saved 100 hours of manual effort by developing Salesforce integration with the payments processor when ACH transactions failed. Re-designed automatic internal disbursement of company funds to reduce manual approval of payment orders by a factor of 100.
Worked on an agile team of 6 member to produce self-serve onboarding and renewal features for health insurance benefits, and create internal tooling for the customer success team. Oversaw onboarding of 3 new engineers, providing contextual and technical support during their first 90 days. Provided engineering support to invoicing and billing, working cross-functionally with the Finance and Business Operations team. Troubleshooted data errors and designed root cause fixes as part of a weekly triage rotation. Improved load time by 200% of a dashboard by consolidating database queries on page render. Re-architected Sana's Modern Treasury API wrapper to improve modularity and test coverage.
Maintained all processing of experimental calcium imaging data on a 60-node bare-metal Kubernetes cluster. Installed and managed Ray Cluster on Kubernetes using Helm to distribute pipeline jobs across the cluster. Refactored pipeline code to distribute jobs on cluster nodes using the Ray API in Python, decreasing runtime by a factor of 12. Implemented deep-learning based multi-modal cell matching of functional and structural neuronal data. Developed software in Python to align neural, behavioral, and stimulus data for multiple new experimental paradigms. Provided technical support and hotfixes to the processing pipeline and in-silico experiments. Designed relational database schemas and tables in MySQL, keeping in mind potential use-cases and users.
As an intern at Second Measure, I worked on two different teams: Data Products and Client Services. As part of the Data Products team, I developed one of Second Measure’s first predictive models to predict per-company quarter-end revenue from observed sales. This project has since made it into the platform to augment the client's investment decisions with dependable forecasts for revenue. As part of the Client Services team, I created a framework and metrics to measure the engagement of a company's customer base based on power user curves. This framework provides potential investors with a way to quantitatively measure how often customers purchase from companies.
I worked as an undergraduate TA for both Calculus 2 and Calculus 3 semester calculus. Every week, I created supplemental curriculum which I went over in four 1-hour long sessions. In addition, I attended lectures and met weekly with the professor to relay feedback from the students.
I designed data-mining and data transformation pipelines for a study on news consumption in the US and Germany funded by the Volkswagen Foundation. Under Dr. Pablo Barbera, I created Web Scraping scripts in R that scraped over 1,000 articles a day from various German news providers (Spiegel, Zeit Online, Frankfurt Allgemeine). As part of the project, I was tasked with creating an R package (using proper commenting, code organization, and packaging) to facilitate further studies on news consumption.