Foundation Model Pre-training Scientist

Kuailu Software; Singapore; On-site; SGD 180,000 - 300,000; Full time

Job summary

Kuailu Software is building a proprietary, LLM-powered Super AI Agent Platform and seeks a Foundation Model Pre-training Scientist to lead our model training strategy.

You will drive pre-training, RLHF/RLAIF, SFT, and architecture decisions, manage distributed GPU infrastructure, and build an AI research team from 0 to 1, aligning foundation model capabilities with product roadmap.

Qualifications

• Hands-on experience in LLM training and foundation models.
• Expertise in pre-training, SFT, RLHF/RLAIF, and optimization.
• Experience with large-scale GPU clusters and distributed training.
• Strong research background in AI/ML, DL, or NLP.
• PhD or equivalent research experience preferred.
• Experience from leading AI labs or top tech companies preferred.

Responsibilities

• Lead Foundation Model/LLM architecture and technology roadmap.
• Drive pre-training, post-training, SFT, RLHF/RLAIF, and model optimization.
• Lead large-scale distributed training and GPU infrastructure.
• Build and lead an AI Research team from 0 to 1.
• Integrate foundation model capabilities into our Super AI Agent Platform.

Skills

LLM training

Foundation models

Model architecture

Distributed training

GPU clusters

AI/ML/NLP research

Education

PhD or equivalent research experience

Job description

We are building a proprietary LLM-powered Super AI Agent Platform and are seeking an Foundation Model Pre-training Scientist to lead our Foundation Model / LLM technology roadmap and model training strategy.

With NVIDIA B300 GPU resources already secured, we are committing dedicated compute and R&D resources to developing our own large-scale foundation model capabilities.

Key Responsibilities

• Lead Foundation Model / LLM architecture and technology roadmap
• Drive pre-training, post-training, SFT, RLHF/RLAIF, and model optimisation
• Lead large-scale distributed training and GPU infrastructure
• Build and lead an AI Research team from 0 to 1
• Integrate foundation model capabilities into our Super AI Agent Platform

Requirements

• Strong hands-on experience in LLM training, Foundation Models, model architecture, or distributed training
• Deep expertise in pre-training, SFT, RLHF/RLAIF, and LLM optimisation
• Experience with large-scale GPU clusters and distributed training
• Strong research background in AI/ML, Deep Learning, NLP, or related fields
• Experience from a leading AI lab, research institution, or top-tier technology company is highly preferred
• PhD or equivalent research experience preferred

This is a research-driven Foundation Model leadership role, not a traditional AI application or Agent product role.