Featured Projects

Predictive Marketing Intelligence Dashboard (Tableau + Machine Learning)

• Designed a fully interactive Tableau system that merges marketing analytics with machine learning–powered foresight, transforming static dashboards into dynamic strategic tools.
• Integrated Random Forest, Gradient Boosting, and Linear Regression models to predict partner performance and contextualize KPIs through Holistic Scoring: a multidimensional metric capturing effectiveness across awareness, spend, and creative mix.
• Built a What-If Scenario Simulator allowing marketers to explore the impact of changing budgets, creative mix, or ML model selection in real time, turning analysis into experimentation.
• Developed a Predictive Leaderboard and Regional Performance Map that visualize over/underperformance, model confidence, and ROI opportunities across markets.
• Demonstrated how AI benchmarks outperform static, retroactive reports by delivering context-aware, adaptive, and actionable insights for budget optimization and media planning. [Read Article on the Dashboard]
  [View Dashboard]

Multi-Real Estate Dynamic Pricing & Market Strategy Simulator (ML + Multi-Agent + GenAI)

• Built an interactive Streamlit app that lets users test “what-if” scenarios for the housing market and see impacts by ZIP code. The engine combines a Ridge pricing model with a lightweight multi-agent market sim (buyers, sellers, analyst) plus GenAI narratives for plain-English insights.
• Data & Quality: Upload CSV or generate synthetic data; schema coercion, basic PII scrubbing, quick EDA (price, sqft, beds).
• Modeling (Ridge): RidgeCV training with error metrics (MAE/RMSE), coefficient explainability, residual diagnostics.
• Scenario Simulator: Adjust interest & unemployment, overlay distributions, ZIP deltas (bar + geo map), save/compare scenarios, shareable URL presets, JSON import/export.
• Agents: Configurable buyer cohorts & seller policies; market clearing over steps; KPIs (price path, volume, sell-through, price-to-ask, time-on-market) with CSV/JSON downloads.
  [View Demo]

Multi-Agent AI Marketing & Business Intelligence System

• Developed a Streamlit-based AI system for a Texas bookstore using orchestrated LLM agents for analytics, segmentation, recommendations, and marketing tasks
• Designed modular agent architecture (BI, Product Expert, RFM Segmentation, Recommendation, Email Writer) with dynamic supervisor routing
• Created an intuitive UI with interactive Plotly dashboards (sales, genre trends, customer segments)
• Built robust preprocessing (e.g., Unicode normalization) and error handling for query stability and system resilience
• Demonstrated end-to-end application of multi-agent systems, NLP, RFM analytics, and operational intelligence
  [View Demo]

AI-Powered Business Intelligence Copilot (Streamlit + LLM + SQL Integration)

• Developed a Business Intelligence Copilot that uses large language models for real-time analytics and decision support
• Integrated OpenAI, LangChain, and SQL to enable natural language querying across multiple databases
• Designed a Streamlit interface with dynamic Plotly visualizations, multi-database routing, and robust error handling
• Extended the system to include a domain-specific (bike sales) database with custom prompts and tailored charts
• Demonstrated applied skills in LLM orchestration, RAG techniques, prompt engineering, and data storytelling
  [View Demo]

RAG Pipeline for Product Knowledge Querying

• Developed an agent-powered app to retrieve and summarize PDFs using LangChain + OpenAI

Testing Framework Design

• Designed 2024/2025 experimentation framework (and all tests) for Chase United paid media (B2C)
• Developed 2022 global experimentation framework for Intel paid media (B2B)
  [View Chase CoBrand Media Case Study] *Due to the sensitive nature of client proprietary information, access is granted upon review
  [View Intel Case Study] *Due to the sensitive nature of client proprietary information, access is granted upon review

Learning Agenda Design

• Introduced Learning Agendas to Quigley Simpson and across Chase CoBrand media (United, IHG, Marriott). Designed framework to rollout learning agendas across Cobrand teams and trained media, analytics and strategy teams on usage
• Introduced and embedded always-on pixel-based Brand Lift measurement across campaigns (Chase, Intel) and trained agency teams on implementation and measurement

• Designed Chase United’s first ever 1^st Party Audience segmentation framework
• Created Intel vPro’s 2021/2022 Learning Agenda
• [View Case Study] *Due to the sensitive nature of client proprietary information, access is granted upon review

Sponsorship ROI Product Roadmap

• Built roadmap and market sizing for Machine Learning + Incrementality + Identity Resolution based solution estimating ROI from brand sponsorships
  [View Case Study] *Due to the sensitive nature of client proprietary information, access is granted upon review

Media Mix Modeling POV

• Authored Merkle’s first ever POV on applying MMM frameworks to B2B marketing strategy
  [Read POV]

RFM Segmentation / K-Means Clustering

• Built Machine Learning-driven customer segmentation solution for an e-commerce client

Natural Disaster Forecasting with Logistic Regression

• Predicted forest fire likelihood using weather-based regression modeling

GOAT Index

• Coined the term in 2015 and built a weighted, indexed NBA player scoring system (new version in development)

Paid Media Partner Indexing for Mercedes-Benz

• Created an indexed scoring system to evaluate media partners across performance metrics — a precursor to regression-based media evaluation
• Approach was adopted for broader partner use

Enterprise Data Flow Mapping for LG Electronics

• Led a cross-functional consulting engagement to document business rules and data flows across all LG business units
• Facilitated executive interviews and team workshops to support the company’s data lake and governance initiative
[View Case Study] *Due to the sensitive nature of client proprietary information, access is granted upon review

Propensity Modeling for AVIS (in partnership with Google Cloud)

• Built a machine learning–driven propensity model to identify high-value AVIS customers and automate marketing activation
• Partnered with Google Cloud engineers to operationalize the solution using BigQuery and Cloud Functions
  [View Case Study] *Due to the sensitive nature of client proprietary information, access is granted upon review