Cortex Applied Research

Pricing Data Science

Self-driving vehicles

Optimization, budgets, modeling, marketing, metrics

Architecture ■ Designed and implemented a scalable architecture to host machine learning models as APIs using Docker containers. This allows other data scientists to easily host production-level models in R & Python, no longer requiring implementation and maintenance from our Data Engineering and DevOps teams.Modeling ■ Built production-level models that:-predict which dealers are likely to cancel subscriptions with us and which features are important to improve relations with them and determine proper pricing.-predict the likelihood that visitors that submitted leads are likely to purchase vehicles. -create pricing models for our products, including a dashboard and easy-to-use input parameters for product owners to adjust pricing weightings.Ad-hoc Web Analytics ■ Work in cross-functional teams performing ad-hoc & exploratory analyses, statistical analysis, and experimental design recommendations. A/B Testing ■ Created a dashboard for A/B testing web metrics for product analytics.Site Performance Alerting ■ System to detect anomalies and breakouts in site performance data and company KPIs.Member of Analytics Development team ■ Automation of many analytics data processes including aggregations of datasets, job scheduling, Cron jobs, Oozie, scripting, Tableau Administration, SQL queries, different databases (Redshift, S3, Databricks etc…)

- Coded Markov chain Monte Carlo (MCMC) algorithms to infer posterior probabilities of fracture parameters.- Used neural networks and random forests to predict subsurface characteristics.- Developed and tested new azimuthal AVO methods on synthetic and physical modeling data.- Performed sensitivity analyses on anisotropic reflection coefficients to be used for fracture analysis.- Modeled frequency-dependent reflections caused by anelastic interfaces.- Incorporated attenuation and dispersion into a finite difference code.- Analyzed noise characteristics and effects in a large multi-stage microseismic data set.- Presented research for CREWES technical talks and sponsor meetings.

- Conducted research on improving azimuthal AVO to infer subsurface properties.- Implemented quantitative interpretation techniques for reservoir characterization.

- Coded a Markov chain Monte Carlo (MCMC) algorithm and applied it to passive seismic data.- Automated downloading and processing of data from online databases.- Estimated the distribution of rupture velocities of earthquakes.

- Developed new seismic imaging techniques that were used to map the lithosphere-asthenosphere boundary (LAB).- Created a new deconvolution method for impedance inversion based on Bayesian inference.- Implemented and compared many deconvolution methods currently used by seismologists.- Used MATLAB to create Markov chain Monte Carlo (MCMC) simulations.- Interpreted deconvolution results to identify geological structure.

- Assisted with the wiring for a new Scanning Tunneling Microscope.- Used LabVIEW to connect and control instruments with a computer.