As part of the Commercial & Investment Bank, JPMorganChase Payments enables organizations of all sizes to execute transactions efficiently and securely, transforming the movement of information, money, and assets. We tackle complex challenges across the payments lifecycle with solutions that facilitate seamless transactions across borders, industries, and platforms. Operating in over 160 countries and handling more than 120 currencies, we are a leading processor of USD payments with daily transaction volumes in the trillions.
As an Applied AI/ML Associate Senior** **within JPMorganChase Payments, you will be a hands-on builder focused on implementing, training, and debugging neural network workflows end-to-end—from data preprocessing through model training, evaluation, and production inference. You will develop models that operate on large-scale document and image inputs (e.g., OCR outputs, scanned documents, forms), and deliver robust ML capabilities under real-world constraints (latency, scale, reliability, explainability, and governance), with special care given to sensitive data.
Job Responsibilities
- Build end-to-end deep learning workflows, including data preprocessing, dataset construction, train/validation/test splits, and leakage prevention checks.
- Implement and debug neural network models (e.g., Transformers, CNNs, LSTMs) for document understanding tasks such as classification, ranking, entity extraction, and language understanding.
- Develop training pipelines, including configuration management, checkpointing, reproducibility, distributed training patterns (where needed), and experiment tracking.
- Create rigorous evaluation harnesses: offline metrics, calibration/thresholding, robustness testing, slice/segmentation analysis, and systematic error analysis.
- Diagnose and resolve training and data issues (label noise, class imbalance, instability, overfitting, data drift, train–serve skew), using targeted experiments and ablations.
- Optimize models for deployment constraints using techniques such as parameter-efficient fine-tuning, distillation, quantization, batching strategies, and latency profiling.
- Contribute to production integration: model packaging, validation tests, inference pipelines (batch and real-time), monitoring/alerting, and rollback procedures.
- Partner with Engineering, Product, and Risk/Compliance to ensure solutions are explainable where required, well-documented, and audit ready.
- Participate in code and model reviews and help raise engineering rigor through best practices and pragmatic standards.
Required Qualifications, Capabilities, and Skills
- Bachelor’s, Master’s, or PhD in Computer Science, Machine Learning, Statistics, Mathematics, or related fields.
- 3+ years of experience building applied ML systems (or advanced degree with equivalent hands-on applied experience).
- Demonstrated experience of training and shipping neural network-based models into production (or production-like) environments and iterating post-launch.
- Strong hands-on experience with deep learning frameworks (PyTorch or TensorFlow) and modern neural architectures.
- Practical understanding of model development fundamentals: preprocessing, training loops, hyperparameter tuning, evaluation design, and failure-mode analysis.
- Familiarity with deploying or integrating models into services/pipelines on AWS (or equivalent cloud) and operating them reliably at scale.
- Strong communication skills, including articulating tradeoffs (accuracy vs latency, false positives vs customer friction, complexity vs maintainability) to technical and non-technical stakeholders.
Preferred Qualifications, Capabilities, and Skills
- Experience with document AI: OCR pipelines, layout-aware NLP, document classification/ranking, entity extraction, or form understanding.
- Experience with fine-tuning approaches (full and parameter-efficient) and model optimization techniques (distillation, quantization-aware approaches, compression).
- Exposure to ML platform/MLOps practices: data validation, model registries, CI/CD for ML, observability/monitoring, and governance workflows.
- Experience with Docker/Kubernetes and modern data platforms (e.g., Databricks, Snowflake) where relevant.
- Exposure to streaming or near-real-time architectures and feature generation, including point-in-time correctness and leakage prevention.
- Experience working with data using SQL and distributed processing (e.g., Spark/PySpark or equivalent).