Machine Learning Researcher - HFT

The team trades at sub-second to intraday horizons across Asian equity markets, where signal decay is fast, capacity is tightly constrained and execution quality is inseparable from alpha. This is a seat for a researcher who can build predictive models under hard latency and microstructure constraints - and see them go live quickly.

Key Responsibilities

• Develop machine learning models for short-horizon price prediction at tick, sub-second and intraday frequencies.
• Research order book dynamics - queue position, order flow imbalance, liquidity provision and adverse selection - and convert them into tradeable signals.
• Engineer features directly from full-depth order book, trade-by-trade and nanosecond-timestamped market data.
• Build models that respect real-world constraints: latency budgets, exchange throttles, fill probability, market impact and transaction costs.
• Own the research lifecycle end to end, from hypothesis through to production deployment and live monitoring of model decay.
• Work in tight partnership with quantitative developers and execution engineers, where research and infrastructure are inseparable.

Requirements

• PhD or Master's in Machine Learning, Statistics, Computer Science, Mathematics, Physics or a related quantitative discipline from a top-tier university.
• Strong applied ML background with genuine understanding of low signal-to-noise, high-frequency data - including online learning, regularisation, and robustness to regime shifts.
• Excellent Python and strong C++ - this is a latency-sensitive environment where research code sits close to production.
• Demonstrated experience with high-frequency market data at scale: L2/L3 order book reconstruction, tick data handling, and clock/timestamp discipline.
• Rigorous approach to backtesting HFT strategies, including realistic fill simulation and slippage modelling.