LLM & Multimodal ML Engineer — E‑Commerce (Seattle, In‑Person)

TikTok USDS Joint Venture LLC; Seattle (WA); On-site; USD 140,000 - 210,000; Full time

Job summary

TikTok USDS Joint Venture LLC in Seattle is seeking a Machine Learning Engineer, E-Commerce to architect and develop advanced LLM-based systems for processing high-dimensional unstructured data from diverse sources.

You will define semantic understanding frameworks, including fine-grained intent classification and entity extraction, and apply techniques such as Instruction Tuning, RLHF, and prompt engineering to improve model performance.

Qualifications

• MS or PhD in Computer Science, AI, Mathematics, or a related field.
• 1+ years of professional experience in one or more of the following fields: Machine Learning, Natural Language Processing (NLP), or Computer Vision.
• Proficiency in software development with at least one of the following programming languages: C++, Python, Go, or Java.

Responsibilities

• Architect and develop sophisticated LLM-based systems to process high-dimensional unstructured data (e.g., text, logs, conversations) from diverse sources.
• Define and implement complex semantic understanding frameworks, including fine-grained intent classification and entity extraction, to achieve deep contextual awareness.
• Apply state-of-the-art techniques to enhance model performance, including Instruction Tuning, RLHF/DPO, and Contextual Prompt Engineering.
• Optimize model serving for production environments using quantization, distillation, and efficient attention mechanisms (e.g., FlashAttention, vLLM).

Skills

Machine Learning

Natural Language Processing

Computer Vision

LLM

Education

MS or PhD in Computer Science, AI, Mathematics

Tools

C++

Python

Go

Java

Job description

TikTok USDS Joint Venture LLC in Seattle is seeking a Machine Learning Engineer, E-Commerce to architect and develop advanced LLM-based systems for processing high-dimensional unstructured data from diverse sources.

You will define semantic understanding frameworks, including fine-grained intent classification and entity extraction, and apply techniques such as Instruction Tuning, RLHF, and prompt engineering to improve model performance.