As a Machine Learning Engineer on the Video AI Platform team, you'll develop and maintain the core systems that power video understanding, summarization, and classification. Your work will center on building scalable pipelines for fine-tuning and retraining models, including LoRA-based workflows and large language models, ensuring they remain accurate and efficient over time.
Key Responsibilities
- Design and manage end-to-end ML pipelines for model training, versioning, and evaluation
- Implement and maintain vector search and semantic similarity systems for content retrieval
- Architect model serving solutions for open-source and foundation models at scale
- Establish RAG pipelines to support video metadata extraction, summarization, and question answering
- Monitor live models for performance drift and trigger retraining when needed
- Optimize compute usage across distributed training environments to control costs
- Integrate human-in-the-loop labeling into training workflows for improved data quality
- Support deployment of diverse model types, including vision-language models and embedding generators
- Build and maintain a feature store with robust data versioning capabilities
- Develop evaluation frameworks to measure LLM accuracy, hallucination rates, and structured output reliability
Qualifications
You bring at least five years of hands-on experience in machine learning engineering, with a strong foundation in full lifecycle ML development. You're skilled in deploying models using tools like Triton, TorchServe, or Ray Serve, and have experience managing distributed compute platforms such as Kubernetes, Ray, or SageMaker.
Proficiency with experiment tracking systems like MLflow or Weights & Biases is essential, as is familiarity with model versioning practices. You have deep knowledge in NLP, recommendation systems, and reinforcement learning, and you thrive in Agile settings while collaborating with global teams.
Experience with real-time inference, streaming data, or human-in-the-loop workflows is a plus. You're comfortable working across the stack to deliver reliable, production-grade AI services that power intelligent video applications.
