Quantitative ML Engineer (PyTorch & PPNR Migration)

Job Title: Quantitative ML Engineer (PyTorch & PPNR Migration)
Location: New York, NY (Hybrid)
Duration/Term: Long Term Contract

Job Description

We are seeking an experienced Quantitative ML Engineer to lead the migration of complex PPNR (Pre-Provision Net Revenue) forecasting models from legacy Hadoop/C++/R environments to a modern Databricks and PyTorch ecosystem. The role involves translating intricate econometric and statistical modeling logic into optimized PyTorch tensor-based implementations while ensuring strict numerical parity for regulatory compliance under CCAR/DFAST frameworks.

The candidate will work closely with quantitative researchers, data engineers, and Model Risk Management teams to modernize regulatory models and ensure scalability, performance, and audit readiness.

Key Responsibilities

• Model Translation: Reverse-engineer legacy C++ and R codebases to extract mathematical logic, econometric formulas, and simulation parameters used in PPNR forecasting models.
• PyTorch Implementation: Rebuild forecasting models using PyTorch, leveraging torch.nn modules, tensor operations, and custom Autograd functions when required.
• Performance Optimization: Optimize models using Databricks distributed computing, GPU acceleration, and efficient tensor operations to significantly reduce execution time.
• Data Integration: Develop scalable pipelines integrating Snowflake data into Databricks using Spark, and create efficient PyTorch DataLoaders for model training and simulation.
• Model Validation: Perform back-testing, sensitivity analysis, and statistical validation to ensure model outputs match legacy Hadoop/Hive-based systems.
• Regulatory Documentation: Work with Model Risk Management (MRM) teams to produce documentation aligned with SR 11-7 regulatory standards covering model architecture, migration methodology, and validation results.

Qualifications

• Master s or PhD in Statistics, Financial Engineering, Mathematics, Physics, Computer Science, or related quantitative field.
• Strong background in quantitative modeling, econometrics, or financial risk modeling.
• Experience working with regulatory financial models or large-scale statistical systems is highly preferred.

Experience

• 6 8 years of experience in Machine Learning Engineering, Quantitative Modeling, or Financial Risk Analytics.
• Hands-on experience developing PyTorch-based models for non-computer vision tasks such as time-series forecasting, regression, or Monte Carlo simulations.
• Proven experience migrating workloads from legacy big-data environments like Hadoop/Hive to modern cloud data platforms.
• Experience working with Databricks, Spark-based ML pipelines, and Snowflake data platforms.
• Familiarity with CCAR, DFAST, or PPNR regulatory models is a strong advantage.

Key Skills

PyTorch, Python, Quantitative Modeling, Econometric Forecasting, Time-Series Modeling, Monte Carlo Simulation, C++, R, Databricks, Apache Spark, Snowflake, Hadoop, Hive, MLflow, Tensor Optimization, PyTorch Autograd, TorchScript, ONNX, Data Pipelines, Financial Risk Modeling, PPNR Modeling, CCAR, DFAST, Model Validation, Backtesting, Sensitivity Analysis, Distributed Computing, GPU Computing, Regulatory Compliance, SR 11-7 Documentation

VDart Group is a global leader in technology, product, and talent solutions, serving clients including Fortune 500 companies across 13 countries. With over 4,000 professionals, we deliver innovation and results across industries. Committed to People, Purpose, and Planet, we are recognized for our sustainable practices through our EcoVadis Bronze Medal and UN Global Compact membership.