Data Scientist

About the Role:

We are seeking Data Scientists with strong expertise in feature inference, feature selection, and model interpretability, leveraging techniques such as Mutual Information (MI), regression, and XGBoost. The role focuses on identifying high-signal features, uncovering linear and non-linear relationships, and developing robust, interpretable predictive models that support critical business decisions.

Ideal candidates will bring real-world, applied AI/ML experience in domains such as unsecured lending, BNPL, insurance, or healthcare. Hudson Data's analytics work spans risk modelling, predictive analytics, and decision intelligence, addressing technically complex problems with significant commercial impact.

At Hudson Data, we bring together business leaders, data scientists, and engineers to transform complex data into measurable business outcomes. Our cross-functional teams combine deep domain expertise with modern data and AI technologies to help clients identify growth opportunities, improve operational performance, and make smarter, faster decisions at scale.

Beyond client engagements, Hudson Data collaborates with academic and industry partners to explore emerging technologies and translate innovation into practical solutions. Alongside our work with Fortune 500 companies, we develop proprietary products designed to address complex and evolving business challenges.

Headquartered in New Delhi, India, with offices in Gurugram, Haryana, and New York, USA, Hudson Data serves clients globally, combining deep technical expertise, domain knowledge, and innovation to deliver lasting business impact.

Key Responsibilities

Feature Inference, Selection & Engineering

• Apply Mutual Information (MI) and statistical techniques to evaluate feature relevance and dependency with target variables.
• Perform feature selection, ranking, engineering, and dimensionality reduction to identify high-signal predictors.
• Identify non-linear relationships and interactions that may not be captured through traditional correlation analysis.
• Assess feature stability, predictive power, and suitability for production models.

Predictive Modeling & Validation

• Develop and evaluate predictive models using Linear and Logistic Regression, Ridge, Lasso, XGBoost, and other gradient-boosting/tree-based techniques.
• Compare model performance across different algorithms and feature sets.
• Conduct cross-validation, hyperparameter tuning, model diagnostics, and performance testing.
• Evaluate models using appropriate metrics, including AUC-ROC, RMSE, Precision, Recall, and related measures.

Model Interpretability & Decision Intelligence

• Analyze feature importance using MI scores, model-based importance, SHAP, and other feature-attribution techniques.
• Explain model behavior, key drivers, and trade-offs to both technical and business stakeholders.
• Translate analytical findings into actionable business insights and decision frameworks.

Data Extraction & Processing

• Write and optimize advanced SQL using complex joins, window functions, aggregations, and transformations.
• Use Python (Pandas, NumPy, Scikit-learn) for data extraction, cleaning, preprocessing, feature engineering, modeling, and validation.
• Work with large and complex datasets while maintaining data quality and analytical accuracy.

Business Applications

Apply predictive modeling and analytical techniques to real-world use cases including:

• Credit risk scoring and payment/default probability
• Fraud detection and risk analytics
• Lead scoring and propensity modeling

Partner closely with business and technology teams to ensure models are aligned with operational KPIs, business objectives, and measurable outcomes.

Required Skills

• Strong proficiency in Python, particularly Pandas, NumPy, and Scikit-learn.
• Advanced SQL skills.
• Hands-on experience with Mutual Information (MI), feature selection, and feature engineering.
• Strong understanding of regression techniques and XGBoost/gradient-boosting models.
• Solid understanding of MI vs. correlation, including linear and non-linear relationships.
• Knowledge of filter, wrapper, and embedded feature-selection methods.
• Strong understanding of model evaluation techniques and metrics, including AUC-ROC, RMSE, Precision and Recall.
• Ability to interpret and explain model outputs, not simply build models.

Good to Have

• Experience with Explainable AI (XAI) and SHAP.
• Feature stability and model-monitoring experience.
• Time-series modeling experience.
• Applied experience in Fintech, unsecured lending, BNPL, credit risk, insurance, healthcare, or fraud analytics.

Education & Certifications

• Bachelors degree in Data Science, Computer Science, Statistics, Mathematics, or a related field.
• Masters degree in Data Analytics, Machine Learning, or Business Intelligence preferred.
• Relevant certifications are highly valued:
• Google Cloud Professional Certifications

Why Join Hudson Data?

At Hudson Data, youll be part of a dynamic, innovative, and globally connected team that combines advanced AI/ML frameworks, modern data technologies, and cloud-based analytics platforms to solve complex, real-world business challenges. You’ll have the opportunity to experiment with new approaches, deepen your technical expertise, and see your work translate into measurable business impact within a culture that values creativity, precision, collaboration, and continuous learning.

Our work spans high-impact domains including unsecured lending, BNPL, insurance, healthcare, and fraud analytics, where data science plays a critical role in solving commercially significant problems. You’ll work on applications involving predictive modeling, risk scoring, fraud detection, customer and payment behavior, and decision intelligence—providing the opportunity to apply advanced analytical techniques to problems where model quality directly influences business outcomes.

For Data Scientists interested in applied machine learning rather than purely theoretical modeling, Hudson Data offers the opportunity to work at the intersection of advanced analytics, domain expertise, and real-world decision-making.