Job Responsibilities
Research and prototype next-generation architectures for structured and unstructured data
Develop novel pre-training objectives tailored to financial event sequences and heterogeneous profile data
Implement research ideas in production-quality code
Mentor engineers on ML best practices; translate research advances into deployable systems
Optimize training throughput for large data sources
Collaborate with other teams to design solutions for product use cases.
Required qualifications, capabilities, and skills
Master’s degree with 2+ years Or Bachelor's with 4+ years in Computer Science, with training and work experience in Machine Learning, LLM/NLP or similar fields.
Deep LLM and Transformer expertise — strong command of attention mechanisms, positional encodings such as RoPE, and the ability to handle multi-modal data inputs effectively.
PyTorch proficiency at scale — hands-on experience with distributed training frameworks including FSDP and DeepSpeed, alongside practical memory optimization techniques.
Foundation model training — proven experience in pre-training from scratch and designing tokens and vocabularies for complex, heterogeneous data sources including tabular, temporal, and graphical formats.
Strong software engineering skills — ability to build robust, production-quality systems that perform reliably at scale.
Prior experience with financial data and recommendation systems.
Preferred qualifications, capabilities, and skills
Publication record at top AI/ML venues.
Experience optimizing serving infrastructure is a plus.
Experience with post-training LLMs and network optimization algorithms, as well as interpretability or steering techniques for LLMs.
Experience working with large-scale compute infrastructure.
Experience shipping a real-world product, project, or feature.
Experimental rigor and ablation design when benchmarking LLM optimizations.
Strong communication and accountability skills, with a collaborative mindset and strong work ethic.
- “This position is subject to Section 19 of the Federal Deposit Insurance Act.”
- “As such, an employment offer for this position is contingent on JPMorganChase’s review of criminal conviction history, including pretrial diversions or program entries.”
