Job
Core
Building and deploying production AI features for a live-streaming platform, balancing the use of frontier LLMs with custom-trained models for tasks like content moderation, recommendation, and fraud detection.
Role type
Senior Machine Learning Engineer (LLM & Model Training)
Builds
AI features for live video/audio streaming, content moderation, recommendation, and creator monetization at scale (50k RPS).
Domain
Live streaming, content moderation, recommendation systems, fraud detection.
Deliverable
production ML models
Required skills
LLM engineering (retrieval, agents, context engineering), model training (SFT, LoRA, distillation), Python, PyTorch, GPU optimization, production ML serving, data pipeline ownership, A/B testing, model monitoring.
Preferred skills
Experience with agentic flows, building evaluation tooling, working with frontier model APIs and open-weight models.
Technologies
Python, PyTorch, LLM APIs, GPU clusters, internal agents, MCP servers.
Responsibilities
Design and implement LLM-based features including context engineering and agentic flows; train and deploy custom models for classification and ranking; manage the full data lifecycle from collection to labeling; serve models under real load with latency and cost constraints; define and execute offline/online evaluation strategies; monitor for drift and regressions in production.
Seniority
Senior, hands-on IC
