about 4 hours ago
Seoul, Korea, SouthStaff+
Responsibilities
- Build and own end-to-end ML pipelines for data, training, evaluation, inference, and deployment.
- Fine-tune and adapt models using state-of-the-art methods like LoRA and QLoRA.
- Architect and operate scalable inference systems balancing latency, cost, and reliability.
- Design and maintain data systems for high-quality training data.
- Implement evaluation pipelines for performance, robustness, safety, and bias.
- Own production deployment, focusing on GPU optimization and scaling policies.
- Collaborate with application engineering to integrate ML systems into products.
- Make pragmatic trade-offs and ship improvements quickly.
Requirements
- Strong background in deep learning and transformer-based architectures.
- Hands-on experience with large-scale ML models in production.
- Proficiency in at least one modern ML framework like PyTorch or JAX.
- Experience with distributed training and inference frameworks.
- Strong software engineering fundamentals for production-grade systems.
- Experience with GPU optimization techniques.
- Comfort with owning ambiguous, zero-to-one ML systems.
Tech Stack
Categories
AI & MLData Engineering