Nicholas Kridler
Director, Machine Learning at Goodwater Capital
Nicholas Kridler
Director, Machine Learning at Goodwater Capital
San Diego, California
Overview
Work Experience
Director, Machine Learning
2022 - Current
Lead a team of data scientists to identify promising startups - Develop KPI models to surface interesting companies - Improved operationalization of ML strategies
Lead Machine Learning Engineer
2021 - 2022
Real-time ML-based Product Recommendation Algorithms - Maintained infrastructure for recommendation systems: real-time ranking models deployed as microservices - Developed and deployed an embedding based personalization module for the e-commerce shop - Provided cost savings via improvements to our dynamodb cache Analytics Platform Support - Developed custom ETLs that retrieved data from ad platforms via REST APIs (e.g. Pinterest) - Developed ETLs for additional data sources in dbt - Provided Looker support to business stakeholders
Senior Data Scientist
2020 - 2021
Led apparel ranking initiatives on the search and personalization team - Delivered a 10% increase to apparel orders through changes to the search algorithms - Developed a process to improve roadmap alignment between the product and ranking teams - Plan Quarterly goals (OKRs) and identify the potential lift in revenue associated with each project Contributed to the core search ranking models - Introduced new features to help characterize predictive behavior of users in search - Train ranking models in interactive (Sagemaker) and batch (Batch) environments - Improve search relevance by designing retrieval algorithms in elasticsearch End-to-End ownership of ranking algorithms - Responsible for backend production algorithm implementation in Kotlin - Develop offline frameworks in Python to evaluate algorithm improvements - Responsible for analyzing A/B tests using variance reduction techniques (e.g. CUPED) to assess the overall impact
Senior Data Scientist
2020 - 2020
Drove adoption of Contextual Bandits - Leveraged ML platform to accelerate the adoption of contextual bandits - Presented proof-of-concept to Senior Leadership Team Influenced product strategy through the development of new metrics - Developed metrics that help characterize customer engagement with Turbo Tax Pioneered new technologies for the team - Led effort to deploy models to the ML platform - Developed tools to enable data scientists and analysts automate their workflows
Lead Machine Learning Engineer
2018 - 2019
Developed product recommendations for the styling service - Leveraged binary classification models for high-dimensionality categorical data - Deployed models through RESTful API microservices Increased sales through recommender model enhancements - Improved recommendations for new customers by tuning the model and training data - Algorithmically combined recommendations to improve stylist efficiency - Evaluated model enhancements through AB tests End-to-End Data Products to support the recommendation platform - Developed dashboards to monitor algorithm performance - Developed metrics to evaluate inventory health Hands-on project management - Weekly one-on-one meetings with team members - Develop roadmap and manage projects via Agile/Scrum
Data Scientist
2014 - 2018
End-to-End ML-based Data Products - Developed a data source used to create ML features that powered multiple capabilities - Built and maintained ETLs using Spark (SQL and PySpark) - Created binary classification models to identify product expansion opportunities - Developed simple web applications to surface recommendations to business partners Drove Vendor Partnership through Sharing Data - Introduced a vendor-facing data portal for sharing performance metrics - Used NLP techniques to characterize customer feedback for each product - Introduced a vendor-buyer collaboration tool for capturing data needed to power other data science initiatives Hands-on project management - Responsible for project scope, implementation, and delivery - New-hire onboarding and mentorship
Senior Data Scientist
2014 - 2014
Transitioned legacy AB Testing Framework code to Python - ETLs and statistical tests to track the results of in-game experiments - Constructed an API to assess potential test biases due to player cohort imbalances Worked towards developing a centralized datastore across all studios - Responsible for defining data model and extracts for the Redshift data warehouse - Collaborated with the engineering team to design aggregate tables Identified the scale and missed revenue opportunity caused by in-game cheating - Created an anomaly detection algorithm to look for abnormal game actions - Presented findings to stakeholders to address revenue loss
Senior Data Scientist
2012 - 2014
Leveraged predictive modeling techniques to improve internal processes - Developed a prioritization algorithm to optimize work-flow Identified potential opportunities for missed revenue using Matrix Factorization - Developed a prototype One-Class Collaborative Filter in R - Successfully identified opportunities not found by human auditors Predicted Inpatient length-of-stay using historical Medicare data - Developed a physician aid for assessing the need for inpatient admittance - Utilized a decade of Medicare claims data
Staff Scientist
2006 - 2012
Apply machine learning techniques to reduce false alarms in detection systems. -Utilize decision trees and support vector machines to discriminate false alarms -Develop additional derived metrics to boost performance Produce end-to-end solutions from data ingestion to actionable insights. -Generate scripts to scrub data and extract additional features -Present results at quarterly meetings Develop production software leveraging GPU image processing in OpenGL. -Implement image filters as GLSL shaders Develop research software for automatic target recognition in OpenCV. -Design C++ software for wide-area template matching
Analyst
2005 - 2006
Conduct Operations Analysis studies using the Naval Simulation System (NSS). Interface with customers to develop and implement new features in NSS. Instruct training classes for new NSS users.