Platinum & Partners is retained by a top-tier global hedge fund to identify an exceptional Machine Learning Research Scientist for a newly formed systematic equities pod. This is one of the most technically ambitious research mandates in quantitative finance right now.
The project: design, implement and train a custom decoder-only Transformer architecture — built from scratch — on tokenised intraday market data for next-token prediction of price movements. This is not fine-tuning an existing LLM. It is a purpose-built architecture for financial time-series, developed with significant dedicated computational resources as part of a long-term research agenda.
PhD-level expertise in deep learning and Transformer architectures is required. The firm wants someone who can start as soon as possible.
THE ROLE
- Design and implement a custom decoder-only Transformer architecture optimised for tokenised financial time-series data
- Develop a novel tokenisation scheme for intraday market data: price movements, volume, order flow and cross-sectional features
- Implement efficient training pipelines in PyTorch — mixed-precision training, gradient checkpointing and multi-GPU parallelism
- Design attention mechanisms adapted to financial data: temporal attention patterns, cross-asset attention and multi-scale representations
- Build evaluation frameworks for next-token prediction accuracy, signal quality and trading performance
- Implement inference optimisation for low-latency production deployment: model quantisation, KV-cache, speculative decoding
- Conduct rigorous ablation studies to validate architecture choices and training methodology
- Collaborate with the team to integrate model predictions into the live trading pipeline
- Define and execute a multi-month research agenda with the autonomy and accountability of a senior scientist
WHAT THEY ARE LOOKING FOR
- PhD in Machine Learning, Computer Science, Statistics, Applied Mathematics or a related field with a deep learning focus — this is a firm requirement
- Demonstrated ability to implement Transformer architectures from scratch — not adapting or fine-tuning existing models
- Deep understanding of attention mechanisms, positional encodings, tokenisation strategies and training dynamics
- Expert-level PyTorch: custom modules, training loops, mixed-precision and multi-GPU training
- Strong mathematical foundations: linear algebra, probability theory, optimisation, information theory
- Experience training models at scale — 100M+ parameters
- Strong Python; C++ for performance-critical components
- Self-directed: capable of defining and executing a complex, long-horizon research agenda with limited supervision
- Published research at top ML venues (NeurIPS, ICML, ICLR) or equivalent industry research experience is a meaningful advantage
- Experience applying deep learning to financial data, time-series forecasting or non-text tokenisation is highly valued
- Knowledge of market microstructure and intraday trading dynamics is a plus
WHY THIS ROLE
- A genuinely rare research mandate — building a purpose-built financial Transformer from first principles, not adapting someone else's architecture
- Significant dedicated computational resources for a long-term research agenda — the firm is fully committed to this project
- A top-tier global hedge fund with the data, infrastructure and capital to support world-class ML research
- Newly formed pod: you are shaping the research direction, not inheriting a fixed approach
- Direct path from research to live trading — your model predictions feed into production
- Base $150,000–$200,000 with a discretionary performance bonus reflecting the firm's Tier 1 position — total compensation is significantly above the base range
- New York City — working alongside the Portfolio Manager and team whose performance depends on what you build
ABOUT PLATINUM & PARTNERS
Platinum & Partners is the specialist quant and systematic executive search firm. Retained and exclusive mandates only. All candidate conversations handled in complete confidence.