DiDi Global logo
DiDi GlobalPosted 1 week ago

Algorithm Expert - Financial Foundation LLM

HybridSan Jose, California, United States

Full TimeEnterprise

Job Summary

Design and implement pre-training pipelines for financial behavior sequence foundation models, selecting objectives like CLM/MLM/Hybrid and experimenting with tokenization architectures. Build unified sequence representations for behavioral data across payments, ride-hailing, and food delivery, while designing an account-card dual-dimension modeling scheme that fuses micro-level event tracking with macro-level transactions. Establish systematic ablation experiment frameworks and freeze-backbone evaluation pipelines to drive architecture decisions, then integrate pre-trained representations into downstream risk-control scenarios including stolen-card detection and credit scoring. Complete SFT fine-tuning and online deployment of blending modules to balance transaction security and user experience.

Required Qualifications

  • Master's degree or above in Computer Science, Mathematics, Statistics, or a related field
  • 3+ years of deep learning algorithm R&D experience
  • hands-on experience building a pre-trained model from scratch
  • completing the full training pipeline
  • Proficiency in Transformer architectures and variants (GPT/BERT/FT-Transformer/MoE)
  • practical sequence modeling experience
  • Familiarity with at least one mainstream deep learning framework (PyTorch preferred)
  • experience with distributed training (multi-GPU / multi-node)
  • Solid experimental design skills
  • ability to independently conduct ablation studies and scaling-law experiments
  • draw reliable conclusions
  • Strong engineering implementation skills
  • able to iterate efficiently on model code and training pipelines

Desired Qualifications

  • Modeling experience in financial risk control / anti-fraud / credit scoring
  • Publications on Foundation Models / Self-Supervised Learning (NeurIPS/ICML/ICLR/KDD/WWW, etc.)
  • Familiarity with Contrastive Learning, ELECTRA, cross-modal fusion, and related techniques
  • Experience with time-series / event-sequence modeling (e.g., TimeMixer, TrajGPT)
  • Experience with graph neural networks or graph-based anomaly detection

Hiring someone like this?

Get your role in front of qualified candidates on Sorce.

Get started

Apply to this job in one click with Sorce

Apply on Sorce