Alvaro Aguado - Data Science Portfolio

About

Hi there! – I’m Alvaro Aguado, a Global Advanced Analytics leader and PhD Data Scientist specializing in AI-driven Analytics, Decision Intelligence, and multimodal machine learning (text + images).

I lead the Global Advanced Analytics group at a large pharmaceutical organization, where I manage marketing effectiveness models, portfolio segmentation, innovation and consumer demand forecasting programs. My work focuses on building scalable data science platforms and explainable AI models that help make faster and data-driven decisions across marketing, supply, commercial strategy, and finance departments.

Over the past 10+ years, I’ve worked across several industries including pharmaceuticals, fintech, and retail, developing data science solutions for organizations such as Sanofi, Pfizer, Deloitte, Google, Yahoo or Verisk. I built transformative capabilities ranging from large-scale data-intensive applications (30B+ rows/month datasets) to real-time analytics leveraging unstructured or limited data sources. My projects include: Marketing Mix Models, Multi-Touch Attribution Models, Online Review + Social Listening Analysis, Influencer and Virality Prediction, Conjoint Models for Innovation, Price Elasticity Analysis, Transfer Demand Models, Segmentation Analysis, Customer Lifetime Value, Computer Vision for manufacturing optimization, among others. More recently I'm working on RAG techniques to improve LLM in the context of insight and decision Intelligence.

In parallel to my work, I completed a PhD in Business Data Science at NJIT (GPA 4.0), where my research focused on how text and images jointly influence online review helpfulness and virality. My work extends the Elaboration Likelihood Model into a quantitative multimodal framework (TI-ELM), explaining how consumers form judgments through informativeness, persuasion, and credibility mechanisms. My research has been published in peer-review outlets including the Journal of Business Research and the Journal of Interactive Marketing.

When I'm not coding I'm taking care of my daughter Livia or playing basketball, tenis or any sport that may seem fun. I also have a passion for dogs and a growing hobby for mixology that's still a work in progress with a lots of errors... Thank you for visiting and I hope to share something you'll find interesting.

Tech stack: Python, R, SQL, Snowflake, TensorFlow/Keras, PyTorch, Bayesian modeling, time-series forecasting using conformal prediction.

Research Interests

01

eXplainable Artificial Intelligence

Methods to understand marketing effectiveness on "black-box" Neural Networks (LSTM, RNN) and tree based ensemble models (catboost, xgboost) vs traditional linear models.

02

Large Language Models

Models applied to Online Reviews and Social Media verbatims. Understanding how unstructured multimodal data can be used for traditional tabulated modeling.

03

Computer Vision

Object detection and information extraction through neural networks (Convolutional Neural Networks) and visual similarity decomposition (p-hash) applied to predict consumer purchase decisions.

04

Time Series Statistical Modeling

Advanced Time-Series forecasting, Conformal Prediction, modern Foundational Models applied to consumer demand, stock market prediction and macro-economic data.

05

Consumer Behavior

Multimodal persuasion and information processing in digital consumer environments, extending the Elaboration Likelihood Model through text–image integration and modeling extreme helpfulness in online reviews.

06

Decision Intelligence & Causal Inference

Marketing Mix Modeling, causal machine learning, price elasticity estimation, and budget optimization frameworks that translate predictive models into strategic decision systems for growth and resource allocation.

Projects

Meridian MMM Platform: Marketing Mix Modeling

Code

A comprehensive end-to-end platform for Marketing Mix Modeling (MMM) built on Google's Meridian framework. This system enables marketing analysts and data scientists to ingest multi-source marketing and sales data, model marketing effectiveness using Bayesian hierarchical models, optimize media budget allocation for maximum ROI, and simulate what-if scenarios for strategic budget planning. The platform includes standardized data ingestion pipelines, prior configuration for Bayesian modeling, budget optimization engines, interactive Streamlit dashboards, and Jupyter notebooks for analysis. Core features include channel effectiveness analysis, ROI estimation, spend optimization with configurable constraints, and PostgreSQL integration for scalable data management. Applied to real-world marketing effectiveness problems with large-scale datasets (>30B rows/month).

Bayesian Modeling Marketing Analytics Python Streamlit PyMC Decision Intelligence

Understanding NLP Algorithms: Independent Study

Code

A comprehensive independent study exploring core Natural Language Processing algorithms and techniques. This project provides hands-on implementation and analysis of fundamental NLP methods, including tokenization, text preprocessing, word embeddings, sequence models, and language understanding techniques. Implemented through Jupyter notebooks with detailed explanations, visualizations, and practical examples. The work demonstrates deep understanding of how NLP algorithms work under the hood, from classical statistical approaches to neural network-based methods, with focus on interpretability and algorithmic foundations.

Natural Language Processing Python Jupyter Notebook Text Analysis Machine Learning Algorithms

Social Tribes & Virality: Twitter Network Analysis

Code Article 1 Article 2 Article 3

An analytical study of social network dynamics on Twitter, examining how communities (tribes) form and connect to drive content virality. This project applies graph analysis techniques to identify influential users, community structures, and network propagation patterns. It uses network science methods to understand how information spreads through social communities and predicts content virality based on network topology and user influence metrics. The work includes advanced graph visualization, clustering algorithms to identify social tribes, and statistical analysis of how network position and community membership affect content amplification and reach on social media platforms.

Graph Analysis Social Network Virality Prediction Python Network Science Community Detection

Blog

Regression & Statistical Modeling

Time Series & Advanced Topics

Get In Touch

I'm always interested in collaborating on interesting data science projects, discussing research, or exploring new opportunities.

Send me an email