Consensus product powered by 🤗 and OpenAILLMs to provide evidence-based answers

Helping companies with data strategy, and execution of best practices in data engineering and data sciencelinear A data products

Transformers to train novel document classifiersScalable automation pipelines to process hundreds of millions of legal documents

- Lead all data strategy, data science, analytics, and data engineering efforts- Data engineering overhaul - migrated to an ELT pipeline to give clean visibility into raw data sources and transformations for final consumable tables (Fivetran & SQL based transformations)- Designed and implemented dashboard tree infrastructure for metric tracking- Prototyped ML - topic modeling for chat data, propensity to re-engage, propensity to save, client segmentation

- Managed data science, analytics, and engineering teams- Data science: responsible for the development and productization of machine learning models- Analytics: managed analytical requests from across the business – statistical analysis and visualizations in Looker- Engineering: focused on ELT pipelines and cloud infrastructure

- Lead a team of 2 data scientists, responsible for ML across the company- Full-stack data science, responsible for both model development and productizing- ML development for sales, marketing and retention: lead and opportunity scoring, churn risk, sentiment analysis, free trial conversion, marketing attribution- Productizing ML to Google Cloud - containerized models on Kubernetes for real-time scoring, Airflow jobs for batch scoring

- Developed machine learning models focused on pricing, revenue growth, deal economics, and deployment patterns- Customer journey model to determine highest probability of upgrading, and highest value customers for AEs to target- Predict customer deployment patterns based on company metadata - use on large quantity purchases to properly ramp pricing- Built polynomial models to quantify volume discounting, and compare price premium between related products- Predicted whether or not each product is on target to meet its growth goal; communicate results to sales teams to properly plan sales strategy- Attributed overall change in average selling price to shifts in quantity and selling price at varying customer sizes; attributed change in revenue to changes in quantity and selling price (Taylor Series Expansion)

- Model developer responsible for the entire modeling process, from data collection and preparation through implementation and on-going monitoring- Built a probability of default (PD) model for CDF Inventory Finance (tier-one model: over $10B in exposure). Generalized linear mixed model with random effects calculated to make adjustments by industry (credibility theory to determine blend between overall model and segmented model). AUC improved by 0.14 and KS improved by 0.13 over the previous implemented model on test set- Built a model to determine the overall risk of the CDF Inventory Finance portfolio. Instead of using PD on a transactional basis, the model determines the overall risk using vector autoregression, with model inputs consisting of current and past economic factors. The model is segmented by industry using mixed model theory- Developed a PD model for origination of new customers using bureau data (tier-one model)