Job Description:

We are seeking an experienced Data Scientist to join our team to develop, optimize, and enhance production-grade machine learning solutions predicting critical operational milestones such as Estimated Start/Boarding Times and Estimated Departure/Completion Times for real-time, customer-facing tracking and journey platforms.

This role requires deep expertise in time-based predictive modeling, scalable data processing within the Databricks ecosystem, and practical familiarity with end-to-end MLOps lifecycle practices.

Key Responsibilities:

Predictive Modeling & Forecasting:

- Design, develop, validate, and optimize machine learning models for key operational milestones and departure/event-time prediction.

- Evaluate regression, gradient-boosting, ensemble, and time-aware modeling approaches based on operational and data requirements.

- Build models resilient to dynamic operating conditions, irregular operational events, and shifting data distributions.

Uncertainty Estimation & Benchmarking:

- Estimate prediction confidence and uncertainty, defining fallback logic for customer-facing prediction stability.

- Continuously benchmark model performance against baselines and operational ground-truth reference timestamps.

Data Analysis & Feature Engineering:

- Analyze historical and near-real-time operational data to uncover factors driving operational timelines including scheduled versus actual milestone progress, delay history, operational status updates, resource/asset allocation, congestion, environmental/external constraints, and movement signals.

- Build reusable feature pipelines and data quality validations.

Databricks Workflows & Scalable Processing:

- Develop scalable data preparation, experimentation, training, and scoring workflows inside Databricks utilizing PySpark, Delta Lake, SQL, and Databricks Workflows.

- Ensure modular, parameterized, reproducible code in close alignment with Data Engineering.

MLOps & Model Lifecycle Management:

- Leverage MLflow for experiment tracking, model comparison, artifact management, and Model Registry governance.

- Define metrics for accuracy, inference latency, data quality, data drift, and model degradation.

- Recommend retraining triggers, validation gates, rollback criteria, and handover specifications for DevOps deployment.

Cross-Functional Collaboration & Delivery:

- Partner with Product, Data Engineering, DevOps, and Operations stakeholders to translate business goals into measurable modeling objectives.

- Support UAT, production rollout, performance monitoring, and deliver thorough technical documentation.

Required Qualifications:

Professional Experience:

- 4+ years of hands-on professional experience in Data Science and Machine Learning, with a track record of developing and delivering production-grade ML models using large structured and semi-structured datasets.

Predictive & Time-Based Modeling:

- Demonstrated expertise in predictive analytics, time-based prediction, forecasting, ETA/ETD estimation, milestone forecasting, or real-time event-time prediction.

Databricks Ecosystem:

- Hands-on experience developing within Databricks, utilizing notebooks, Delta Lake, Databricks Workflows, and Spark-based processing (PySpark).

- Unity Catalog exposure is preferred.

MLOps Lifecycle Awareness:

- Practical experience with MLflow (experiment tracking, model registry, artefact versioning), reproducible pipeline design, CI/CD concepts, drift detection, and automated retraining/rollback workflows.

Programming & Core Languages:

- Advanced proficiency in Python, SQL, and PySpark for scalable data manipulation, feature engineering, and statistical analysis.

Evaluation & Validation:

- Solid grounding in evaluation metrics including MAE, RMSE, percentile errors, threshold-based accuracy, calibration, segment-level validation, and temporal back-testing.

Cross-Functional Collaboration:

- Proven ability to work in Agile environments collaborating closely with Data Engineering, DevOps, and business stakeholders.

Education:

- Bachelor's or Master's degree in Data Science, Computer Science, Statistics, Mathematics, Operations Research, Artificial Intelligence, or a related quantitative field.

Preferred Qualifications:

Domain Experience:

- Prior experience in transportation, mobility, logistics, supply chain, or similar real-time event-time and milestone prediction systems.

High-Reliability Customer Products:

- Experience delivering customer-facing prediction products with strict requirements for low latency, high stability, and explainability.

Cloud & Lakehouse Architecture:

- Experience working across enterprise cloud environments and modern lakehouse data architectures.