Machine Leaming Engineer - Applied ML for Trading Signals

Millennium•New York, NY

8d•$150,000 - $200,000•Onsite

About The Position

We are seeking an applied ML engineer to develop, optimize, and deploy machine learning models for alpha generation within a newly formed systematic equities pod deploying intraday mean reversion and microstructure strategies. The focus is on tree-based ensemble methods (LightGBM, XGBoost, CatBoost) and classical ML pipelines applied to high-frequency financial data. You will work closely with the Portfolio Manager and quantitative researchers to turn ML models into live trading signals.

Requirements

Master's degree in Computer Science, Statistics, Mathematics, Machine Learning, or a related quantitative field
3+ years of experience building and deploying ML models in a production environment, preferably in finance
Deep expertise in tree-based ensemble methods: LightGBM, XGBoost, CatBoost including hyperparameter tuning, regularization, and feature selection
Strong programming skills in Python with proficiency in scikit-learn, Polars/Pandas, NumPy
Strong understanding of overfitting, data leakage, and proper evaluation methodology for financial time-series
Strong analytical thinking, attention to detail, and intellectual curiosity
Excellent communication skills and ability to explain model behavior to non-ML stakeholders
Familiarity with AI-assisted development tools (Cursor, Claude Code)

Nice To Haves

Experience with financial market data (tick data, order book, corporate actions)
Knowledge of market microstructure and intraday trading dynamics

Responsibilities

Develop and optimize tree-based ensemble models (LightGBM, XGBoost, CatBoost) for intraday alpha prediction
Design and implement end-to-end ML pipelines: feature engineering, training, validation, deployment, and monitoring
Build robust cross-validation frameworks adapted to financial time-series (purged k-fold, walk-forward)
Engineer features from market microstructure data: order flow imbalance, spread dynamics, volume patterns, cross-asset signals
Implement model explainability tools (SHAP, feature importance) to understand and validate signal sources
Optimize model inference for low-latency production deployment
Monitor model performance in production: detect drift, staleness, and regime changes
Collaborate with the C++ developer to integrate ML predictions into the real-time trading engine
Experiment with TabPFN and other rapid-prototyping tools for fast signal discovery