Josh Scherbenski

Founded and built a consumer digital health platform owning product, engineering, data, analytics, and AI systems end-to-end

• Built HIPAA-compliant ML and data infrastructure powering personalized health recommendations for 1,000+ users
• Implemented LLM parsing and agentic workflows to turn medical research, user-reported outcomes, and longitudinal health signals into structured recommendations
• Designed production data pipelines, analytics tracking, experimentation, and evaluation workflows for personalization, responsible AI, and model iteration

Architected the firm’s first AI-powered sourcing and investment intelligence platform, translating structured company data, founder signals, financial indicators, and unstructured text into predictive analytics workflows

• Built ensemble models using Random Forest, XGBoost, regression, and domain-specific features to evaluate 110,000+ companies, increasing qualified deal flow 4×
• Engineered modeling pipelines using 300+ behavioral, financial, and founder-quality parameters, with cross-validation and out-of-time testing for generalization
• Created Snowflake + dbt + Airflow pipelines supporting LLM-based summarization, predictive scoring, ranking, analytics, and model evaluation

Partnered with Marketing, Sales, Product, and Engineering to build analytics, experimentation, and data infrastructure for a SaaS prop-tech platform

• Designed automated attribution, real-time ETL, and ML targeting pipelines increasing new-user acquisition 50%
• Built causal inference models to measure channel incrementality, improving CAC/LTV 25%
• Developed forecasting, churn, funnel diagnostics, and retention analytics to improve growth and product-market-fit decisions

Led a global 33-person organization across data science, ML engineering, pricing, analytics, and data platforms supporting a $100M+ marketplace business.

• Built dynamic-pricing ML models generating 2M+ optimized daily price points and increasing gross profit $10M/year
• Developed geospatial regression and demand-surface models supporting underwriting and market risk decisions, reducing launch time 3 months
• Built centralized data lake and MLOps platform enabling 10+ production ML models
• Established model monitoring, retraining, governance, and operational workflows for production ML reliability

Developed NLP automation cutting manual processing costs 50%, enabling profitable unit economics

• Defined the organization’s first KPI framework and statistical monitoring systems

First data scientist at Glassdoor; architected experimentation, ML, and analytics foundations for marketing, product, and international growth.

• Created job-search experimentation framework using bootstrap, nonparametric, and Bayesian methods
• Built semi-supervised models to predict user/customer value and optimize paid marketing and email sequencing
• Led 8-person data science team; supported product and international expansion contributing to 15M+ new users

focused on probabilistic modeling, systems engineering, and innovation programs