Principal AI Researcher (Post-Training Foundation Models)

Principal AI Researcher responsible for post-training foundation LLMs used in quantitative research and trading workflows. Duties include customizing and fine-tuning foundation models on proprietary datasets for financial language and structured/tabular data; building agentic reasoning and tool-use workflows; developing end-to-end training and evaluation pipelines focused on reasoning, quantitative performance, and factual accuracy; applying alignment techniques such as RLHF and DPO; deploying retrieval-augmented generation systems connecting in-house models to broader knowledge sources; and driving production improvements to reduce inference latency while integrating solutions into live trading and research environments. Must demonstrate deep understanding of transformer internals, post-training adaptation (supervised fine-tuning, LoRA, preference-based alignment), orchestration frameworks and reasoning strategies, and systems-level fluency for GPU memory management and distributed training. Strong Python and ML tooling experience (PyTorch, JAX, Hugging Face, DeepSpeed) and a finance-centric mathematical foundation. PhD or Master’s in a related field; finance domain experience is helpful but not required; relocation is covered.