Tomomi’s AI and Machine Learning Portfolio

AI systems do not operate in isolation — they shape how people decide, trust, behave, and interact.

This portfolio explores machine learning, user behavior analytics, AI safety, and generative AI through practical, real-world applications.

I’m Tomomi Tanaka, an economist and technical program leader working at the intersection of AI, behavioral science, safety, and large-scale decision systems. My work spans machine learning, experimentation, user behavior analytics, AI safety, and generative AI evaluation across technology platforms, digital products, and public-sector systems.

Topics Covered

• Machine Learning with Python
• User Behavior Analytics with BigQuery ML
• Predictive Modeling and Experimentation
• AI Safety and Responsible AI
• Generative AI Safety
• Explainability and Robustness
• Human Behavior and Decision Systems

Generative AI for Data Analysis

This section explores how large language models (LLMs) such as GPT-4, Claude, and Gemini are transforming modern data analysis and research workflows. Beyond conversational applications, generative AI models are increasingly being used as powerful analytical tools for classification, evaluation, pattern detection, and insight generation across complex datasets.

Through practical case studies and research-driven projects, this series examines how generative AI can support scalable, data-driven analysis while also revealing the strengths, limitations, and risks of current AI systems.

GenAI vs Crypto Scammers: Which LLM Wins

Topics Covered

• LLM-Based Data Analysis and Classification
• Comparative Evaluation of Multiple LLMs
• Prompt Engineering and Evaluation Design
• Multilingual and Cross-Cultural Analysis
• AI-Assisted Text Classification
• Behavioral Pattern Detection
• Generative AI Benchmarking
• Research Methodology and Validation
• Ethical Considerations in AI-Assisted Analysis
• Real-World Applications of Generative AI in Research

Areas of Focus

Readers will learn how to:

• Design evaluation frameworks for comparing LLM performance
• Build secure and unbiased AI testing pipelines
• Use generative AI models for large-scale text analysis
• Analyze multilingual datasets and cultural communication patterns
• Evaluate model strengths, weaknesses, and failure cases
• Apply rigorous research methodologies to AI-assisted analysis

Real-World Applications

This section includes practical projects and case studies demonstrating how generative AI can be applied to real-world analytical challenges.

Examples include:

• Scam and fraud detection using LLMs
• Behavioral pattern analysis in online conversations
• Cross-model benchmarking and evaluation
• Multilingual classification and cultural analysis
• AI-assisted research workflows and experimentation

Each project combines technical implementation, empirical evaluation, and real-world context to demonstrate how generative AI can support modern data science while maintaining scientific rigor, transparency, and responsible AI practices.

Generative AI Safety

In this section, I explore the critical realm of Generative AI safety, addressing the challenges and opportunities presented by this revolutionary technology. From ChatGPT’s conversational abilities to DALL-E’s artistic creations, generative AI is reshaping our world. This series aims to equip you with the knowledge to navigate the complex landscape of generative AI safety.

Key topics include:

1. Introduction to Generative AI and Safety Concerns
2. Case Studies
3. Ethical Considerations in Generative AI
4. Misinformation and Deepfakes
5. Content Moderation and Filtering
6. Adversarial Attacks
7. Alignment and Control
8. Transparency and Explainability
9. Safety in Language Models
10. Image and Video Generation Safety
11. Safety in Creative Applications
12. Human-AI Collaboration in Safety

Readers will learn how to:

• Understand key risks associated with generative AI systems
• Evaluate safety challenges in language and image generation models
• Analyze misinformation, manipulation, and abuse scenarios
• Explore alignment and oversight approaches for advanced AI systems
• Design safer human-AI interaction workflows
• Examine the trade-offs between innovation, usability, and safety

The series combines technical concepts, practical implementation strategies, and real-world examples to provide a comprehensive introduction to the rapidly evolving field of generative AI safety.

AI System Safety

This section focuses on the safety, reliability, robustness, and responsible deployment of machine learning systems. As AI systems become increasingly integrated into high-impact products and decision-making environments, understanding how to evaluate and mitigate risks is becoming essential for both technical and non-technical practitioners.

Through practical Python implementations and real-world examples, this series explores key techniques for improving the fairness, transparency, robustness, and governance of machine learning models.

Topics include:

1. Fairness, Bias Detection and Mitigation
   • AIF360 library
   • Reweighing
   • Disparate Impact Analysis
2. Model Explainability and Interpretability
   • SHAP
   • LIME
   • Partial Dependence Plots (PDP)
3. Reliability and Robustness
   • Adversarial training
   • Robust Model Evaluation
4. Ethical Considerations
   • Calibrated Equalized Odds
   • Mitigation
5. Adversarial Robustness
   • PGD attack
   • Adversarial Training with PGD
   • TRADES
   • Randomized Smoothing
6. Privacy-Preserving Machine Learning
7. Scalable Oversight of AI Systems
   • Recursive Reward Modeling
   • Debate and Amplification Techniques
   • Factored Cognition Approaches
   • Human-AI Interaction Protocols
8. Ethical AI Development
   • AI Development Lifecycle

This series emphasizes both the technical implementation and the broader real-world implications of machine learning safety, helping readers better understand how to build and evaluate AI systems that are more trustworthy, transparent, and resilient.

User Behavior Analytics with BigQuery ML

This series explores how to analyze user behavior and build predictive models using SQL and BigQuery ML. Designed for data analysts, marketers, product teams, and business intelligence professionals, the series focuses on practical approaches to transforming large-scale behavioral data into actionable business insights.

Using BigQuery’s scalable analytics environment, readers will learn how to analyze customer interactions, measure engagement, and develop machine learning models directly within SQL workflows.

What You’ll Learn

The series covers a wide range of user behavior analytics and predictive modeling techniques commonly used in digital products and e-commerce platforms.

Topics include:

1. Analyzing User Behavior on an E-commerce Site
2. Deep Dive into User Engagement Analysis
3. Sales Prediction
   • Logistic Regression
   • Random Forest
   • XGBoost
   • Deep Neural Network (DNN)
4. Revenue Prediction
   • Linear Regression
   • Ridge Regression
   • Lasso Regression
   • Random Forest
5. Identifying High-Value Customers
   • Logistic Regression
   • K-Means Clustering
   • Random Forest
6. Customer Segmentation
   • K-Means Clustering
   • PCA + K-Means Clustering
7. Predicting User Conversion
   • Logistic Regression Model
   • Random Forest Model
   • XGBoost Model
8. Churn Prediction
   • Logistic Regression Model
   • Random Forest Model
   • XGBoost Model
9. Recommendation and Personalization
   • Matrix Factorization Model
10. Optimizing Marketing Campaigns
    • Logistic Regression
    • Random Forest
    • XGBoost
    • Deep Neural Networks (DNN)

Each post combines practical SQL implementations, machine learning workflows, and real-world business use cases to demonstrate how behavioral analytics can support product, marketing, and strategic decision-making.

User Behavior Analytics with Python

This section explores the emerging challenges and opportunities associated with generative AI systems, including large language models, image generation models, and multimodal AI applications. As generative AI becomes increasingly integrated into products, platforms, and everyday workflows, understanding how to evaluate and mitigate safety risks is becoming critically important.

Through practical examples, case studies, and technical discussions, this series examines the safety, reliability, governance, and societal implications of generative AI technologies.

Topics covered:

1. Data Cleaning
2. Feature Engineering
3. Sales Prediction
4. Revenue Prediction

The series combines practical Python implementations with real-world analytics use cases to demonstrate how machine learning can support product, marketing, and decision-making strategies.

Price Prediction with Python

This series explores practical machine learning techniques for predicting house prices using Python and the popular Kaggle dataset, House Prices – Advanced Regression Techniques.

Rather than focusing solely on prediction accuracy, this series emphasizes building models that are interpretable, explainable, and reliable in real-world settings. Through hands-on examples, readers will learn how to design end-to-end machine learning workflows while understanding the trade-offs between model performance, complexity, and transparency.

What You’ll Learn

The series walks through the full machine learning pipeline, from raw data preparation to advanced ensemble modeling and model evaluation.

Topics include:

1. Data Cleaning and Preparation
2. Visualization
3. Feature Engineering
4. Feature Selection
5. Model Implementations
   • Linear Models (OLS, Ridge, Elastic Net Regressions)
   • Random Forest
   • Gradient Boosting (GBDT, XGBoost, LightGBM)
   • Support Vector Machines
   • Neural Networks
6. Cross Validation
7. Hyperparameter Tuning
8. Ensemble Learning
9. Interpretability and Explainability

Each post includes practical Python implementations, detailed explanations of core concepts, and links to full GitHub repositories to support hands-on learning and experimentation.

Let’s Connect

I’m always interested in discussions and collaborations related to AI, behavioral science, machine learning, digital safety, and decision systems.

Whether you’re exploring new ideas, building AI-driven products, conducting research, or interested in potential collaborations, feel free to reach out.