Lead Machine Learning Engineer (Foundation Models)
On-siteSingapore, Singapore
Job Summary
Architect pre-training pipelines for large language and multimodal foundation models, orchestrating efficient, scalable workflows from raw data curation through converged checkpoints. Build distributed training systems across multi-node GPU clusters using advanced parallelism strategies like FSDP, DeepSpeed, and Megatron-LM. Design and deploy generative recommendation systems that unify retrieval and ranking against marketplace objectives through post-training loops including SFT, distillation, and preference alignment. Optimize transformer architectures tailored for business use cases, including Mixture-of-Experts and long-context attention. Lead code reviews and mentor junior engineers to sustain production-grade codebases while partnering with cross-functional teams to integrate custom models into live environments. Report to the Senior Machine Learning Engineering Manager onsite in Singapore One-North.
Required Qualifications
- At least 8 years of professional experience in machine learning
- Deep focus on deep learning
- Transformer-based architectures
- Software development lifecycle
- Proven hands-on track record of pre-training or continually pre-training open foundation models (e.g., Llama, Qwen, DeepSeek, Mistral) from scratch
- Deep technical proficiency with multi-node, multi-GPU scaling frameworks such as PyTorch FSDP, DeepSpeed, Megatron-LM, and Ray
- Understanding of modern hardware accelerators
- Practical experience in LLM post-training methodologies, including Supervised Fine-Tuning (SFT), Parameter-Efficient Fine-Tuning (LoRA/QLoRA), and preference alignment methods (RLHF, DPO, RLAIF)
- High proficiency in writing clean, maintainable, and testable production code in Python/C++
- Solid experience in MLOps/LLMOps deployment pipelines
- Experience engaging with AI tools and emerging technologies (such as LLM assistants, code generators, and specialized developer agents) to enhance personal productivity, optimize engineering workflows, and contribute innovative platform ideas
- Experience onsite in our Singapore One-North headquarters
Desired Qualifications
- Exposure to generative retrieval
- Semantic tokenization
- Sequence modelling of user behaviour
- Unifying retrieval and ranking with preference-aligned models
Hiring someone like this?
Get your role in front of qualified candidates on Sorce.