Data Scientist

$30-$70/ USD hour

Part Time

Remote

About the role & the client

We are building an advanced credit scoring system for international newcomers who lack domestic credit history. Our mission is to combine financial behavior data, banking transactions, demographic information, and behavioral patterns to create a fair and practical credit assessment model.

To accelerate this initiative, we are seeking a high-level Data Scientist with deep expertise in statistics, machine learning, and financial analytics to join us on a part-time basis. You will work hands-on across the full modeling pipeline—from data preparation to production-ready model design—focusing on both predictive accuracy and explainability.

Key Responsibilities

1. Data Preparation & Analysis

• Clean, validate, and structure financial transaction data, demographic data, and behavioral data
• Handle missing values, outliers, and data inconsistencies
• Construct training datasets based on defined delinquency and default labels

2. Feature Engineering

• Design features that reflect the unique financial patterns of international newcomers
• Build time-series and behavior-based aggregated features
• Implement reproducible feature generation using Python and SQL

3. Model Development & Evaluation

• Develop baseline statistical models (e.g., Logistic Regression, Ordered Logistic Regression)
• Build and refine machine learning models (Random Forest, Gradient Boosting, Neural Networks, etc.)
• Evaluate models using AUC, KS, ROC, calibration metrics, and other indicators
• Conduct model interpretation using SHAP or other explainability tools

4. Fairness & Explainability

• Perform bias testing across demographic attributes (e.g., nationality, age)
• Document modeling assumptions, limitations, and risk considerations
• Support the development of credit policy guidelines

5. Production Integration

• Draft specifications for integrating models into APIs or batch scoring systems
• Contribute to model monitoring, retraining pipelines, and drift detection
• Create dashboards and BI outputs using Tableau or similar tools

Required Qualifications

Candidates must meet all of the following requirements:

Statistical & Analytical Skills

• Graduate-level understanding of statistics, probability, and linear models
• Hands-on experience with binary classification model development

Programming

• Strong proficiency in Python for data analysis
• (pandas, scikit-learn, TensorFlow, statsmodels, etc.)
• Solid SQL experience for data extraction and manipulation
• Experience using GitHub for code management and collaboration

Domain Knowledge

• Practical experience in financial analytics, credit risk, insurance, investments, or ESG-related data work
• Ability to explain model findings and translate insights into business actions

Communication

• Japanese: business level (documentation, internal meetings)
• English: ability to read technical materials (verbal communication a plus)

Preferred Qualifications

• 3 year + Experience developing credit risk or scoring models (banking, credit card, fintech, etc.)
• Experience analyzing large-scale datasets (100GB+) or healthcare/financial big data
• Experience with AWS (S3, Athena, Lambda, SageMaker, etc.)
• Dashboard building (Tableau, Looker Studio)
• Experience with VBA, R, or MATLAB
• Background in risk modeling, derivatives, or project finance
• Practical experience in at least one of the following:
• Time Series Analysis
• Bayesian Statistics
• Causal Inference
• Unsupervised Learning

Who We’re Looking For

• Strong motivation to solve the social challenge of “credit invisibility” among newcomers
• Commitment to fairness, transparency, and responsible AI
• Comfortable working in a zero-to-one product environment
• Able to collaborate across product, engineering, and business teams

Work Conditions

• Expected workload: 10–20 hours per week
• Weekly online meetings: 1–2 sessions
• Contract: initial 3-month term with possible extension