Machine Learning Engineer

Shields Group Search; New York (NY); Hybrid; USD 140,000 - 210,000; Full time

Job summary

Shields Group Search is seeking an ML Engineer for its Advanced Computing team in a hybrid Manhattan setting. You will own the predictive and optimization layer across the platform, from research to production.

The role covers time-series forecasting, yield optimization, and liquidity forecasting, with end-to-end ownership and no dedicated MLOps team, reporting to Head of Advanced Computing. Strong experience in ML operations and financial data is required.

Qualifications

• Time Series Forecasting: 3+ years applying ARIMA, LSTM, Prophet, or TFT to forecasting with measurable accuracy.
• Optimization: Constrained optimization in production (LP/MIP) for portfolios.
• MLOps & CI/CD: end-to-end ML ops, model versioning and monitoring.
• AWS ML Stack: SageMaker, Lambda, DynamoDB, S3, event-driven architectures.

Responsibilities

• Own the three domain tracks: forecasting, yield optimization, liquidity forecasting.
• Develop end-to-end predictive models and deploy with observability and monitoring.
• Collaborate to implement production-grade conversational interfaces for advisors.

Skills

Time Series Forecasting

Optimization Algorithms

MLOps CI/CD

AWS ML Stack

Containerization

Python Engineering

Tools

SageMaker

Lambda

DynamoDB

S3

Docker

Kubernetes

TensorFlow / PyTorch

Job description

Location: Hybrid, Manhattan, New York

ML experience in fintech / financial services / banking is a major plus

About the Company

Our client is an open finance platform serving wealth managers. The platform connects data from custodians, market-data providers, and other sources with bank and non-bank lenders to facilitate the free flow of capital across financial institutions and their clients.

Built for the next generation of private banking, the platform is data-rich, highly secure, and designed to scale.

About the Role

We are hiring an ML Engineer to join the Advanced Computing team and own the predictive and optimization layer that drives intelligence across the platform.

Your core scope spans three domains:

• Time-series forecasting
• Yield optimization across diversified client portfolios, helping clients earn more on idle cash and reduce cost of debt
• Deal-complexity prediction surfaced through a production-grade conversational interface built for financial advisors

Ownership is end-to-end. From model research and experimentation through deployment, observability, and ongoing monitoring, you will close the loop yourself. There is no dedicated MLOps or platform engineering team between you and production.

This role reports to the Head of Advanced Computing.

Responsibilities

All three tracks are delivered through a unified agentic framework. You will design the agents, own the models they call, and maintain the full stack in production.

Deal Complexity Prediction

• Own the deal-complexity modeling track end-to-end. Other Advanced Computing roles may contribute validation data, but this role is the sole owner of the model and its deployment.
• Train models predicting deal complexity on a 1-10 scale and time-to-close from deal attributes including size, collateral type, party count, and client history.
• Surface predictions through advisor-facing agents to prioritize pipeline and flag high-risk deals early.

Yield Optimization

• Build constrained optimization strategies across a diversified set of client portfolios.
• Own the conversational engine delivering cash-drag analysis, sweep recommendations, and credit-utilization guidance.
• Design client-approved automated execution flows and integrate with banking APIs for real-time balance data and transaction execution.

Liquidity Forecasting

• Build 30/60/90-day cash-position models, including TFT for seasonal patterns and TimesFM for multivariate forecasting.
• Implement event detection across recurring flows, tax schedules, and private-equity capital calls; auto-trigger lending flows on predicted shortfalls.
• Deliver scenario-planning and stress-testing tools for advisor use.

Technical Requirements

Must-Have

• Time Series Forecasting: 3+ years applying ARIMA, LSTM, Prophet, or TFT to real-world forecasting problems with measurable accuracy outcomes.
• Optimization Algorithms: Constrained optimization in production, including linear programming, mixed-integer programming, or genetic algorithms applied to resource-allocation or portfolio problems.
• MLOps & CI/CD: End-to-end ML operationalization, including CI/CD pipelines for model testing and deployment, automated retraining triggers, model versioning, and data and concept-drift monitoring as first-class production concerns.
• AWS ML Stack: SageMaker, Lambda with Python, DynamoDB, S3, and event-driven architectures including SQS and MSK.
• Containerization: Docker for reproducible packaging and Kubernetes/EKS for scalable inference workloads.
• Python Engineering: pandas, numpy, scikit-learn, statsmodels, TensorFlow, and/or PyTorch.

Highly Desirable

• Financial-domain experience in wealth management, treasury, liquidity management, or portfolio optimization.
• Experience building real-time systems, including low-latency prediction APIs for financial decision support with sub-three-minute p95 performance.
• NLP/document-processing experience with AWS Textract, Bedrock including Claude/Titan, and entity extraction from unstructured financial documents.
• Advanced optimization experience, including quantum annealing, D-Wave, or AWS Braket.

This is an opportunity to help architect a state-of-the-art platform and partner with an experienced leadership team during a period of rapid development.