6 months ago
Palo Alto, CA, USASenior
Responsibilities
- Design, build, and maintain end-to-end ML systems, including synthetic data pipelines, model training, debugging, and performance evaluation.
- Fine-tune large language models and implement meta-learning methods to improve model generalization and efficiency.
- Improve existing models by incorporating advancements from recent machine learning research.
- Architect robust model-centric systems and integrate them into real-world applications.
- Scale inference infrastructure for LLMs and VLMs and develop reliable, maintainable AI-driven features.
Requirements
- Hands-on experience training and fine-tuning LLMs and VLMs, including pre-training, instruction tuning, and alignment techniques such as GRPO, RLHF, DPO, and PPO.
- Experience with deep learning models, especially Transformers.
- Ability to translate research papers into clean, production-ready code.
- Experience scaling LLM/VLM inference infrastructure using model-serving frameworks such as vLLM and TGI.
- Proficiency in Python with a strong record of building substantial projects.
- Strong computer science fundamentals, including data structures, algorithms, and design patterns.
- Bachelor’s degree in computer science or a related technical field.
- Experience with PyTorch and TensorFlow.
- Preferred qualifications include an MS or PhD, familiarity with Spark and Airflow, multi-node GPU training experience, open-source ML contributions, and knowledge of Linear Programming.
- Preferred experience includes advanced NLP, multimodal post-training, model distillation, quantization, deployment optimization, and LLM/VLM serving optimization.
- Hands-on experience with AWQ, GPTQ, FP8, or GGUF quantization techniques.
Tech Stack
Categories
About Nace AI
Nace AI builds enterprise AI agents and long-horizon reasoning models that automate financial audit, accounting, and compliance workflows. Its flagship offering includes NAVI for real-time insights across audit, compliance, and operational intelligence, delivered as enterprise software to large organizations. Founded in 2024 and headquartered in Palo Alto, the company is backed by General Catalyst, Walden Catalyst, and Intel.