Lead Engineer / Engineer, AI R&D (LLM), Q Team CoE

The Home Team Science and Technology Agency (HTX) is a statutory board under the Ministry of Home Affairs (MHA) which aims to pioneer innovation solutions and develop world class science and technology capabilities to transform and empower the Home Team in delivering safety and security for Singapore. You will join the core team behind the Phoenix LLM – a family of Sovereign AI models trained right here in Singapore. We are an ambitious and dynamic team of Applied AI Scientists building state-of-the-art Large Language Models (LLMs) and Large Multimodal Models (LMMs) from the ground up. In this role, you will have the unique opportunity to push the frontier of Sovereign AI while collaborating on joint Applied R&D projects with global AI labs. You will have access to large GPU clusters for distributed AI training More about our work here: http://go.gov.sg/phoenix-launch and https://arxiv.org/abs/2605.10391 About the role: • Design and implement scalable architectures for pre-training and post-training large-scale LLMs and LMMs on distributed GPU clusters • Architect data curation pipelines, including data curation and synthetic data generation, for textual and multimodal training datasets • Develop custom evaluation metrics and automated pipelines to assess model performance and alignment • Leverage SLURM for efficient job scheduling and compute resource management in high-performance computing environments • Implement advanced model optimisation techniques, including mixed-precision training and model parallelism • Partner with cross-functional teams to integrate foundational models into agentic workflows and downstream applications • Co-author research publications and collaborate closely with academic and industry partners to push the frontier of Sovereign AI You may specialise in one or more areas in the team: • Data: Build scalable data curation pipelines for large-scale textual, multimodal, and agentic training datasets. Perform data ablations, build synthetic generation engines, and design novel curation techniques • Training: Build robust distributed training pipelines. Run pre-training and post-training (e.g., SFT, DPO, GRPO) experiments for LLMs / LMMs, and design large scale Reinforcement Learning frameworks for agentic model development • Evaluation: Design custom LLM/LMM evaluation metrics. Build and integrate continuous evaluation pipelines into the training framework for testing model checkpoints • Alignment and Constitution: Develop frameworks for model alignment and safety. Research scalable oversight, constitutional AI, and preference tuning methodologies to align agentic models with human intent • Science: Drive foundational research on emerging topics, such as domain adaptation, language transfer, efficient AI architectures, and constitutional agentic AI You may be a good fit if you: • Hold minimally a Bachelor’s degree in Computer Science, Engineering, Mathematics, or a related field. Master’s and PhDs are a strong plus (and well-represented on our team), but exceptional candidates from non-traditional backgrounds with strong portfolios are equally encouraged to apply • Are proficient in deep learning and optimization frameworks, particularly PyTorch, DeepSpeed, and Megatron-LM • Have a proven track record in post-training and evaluating foundational language models. (Experience with reinforcement learning and agentic model development is a strong plus) • Possess a deep understanding of core machine learning principles and modern LLM development best practices (exceptional candidates have hands on experience with SLURM and distributed GPU training) • Bring excellent problem-solving skills and a low-ego, highly collaborative mindset to the team • Bonus points if you have contributed to open-source ML projects or published in AI/ML conferences (main, workshop, poster, or industry tracks are all welcome) **** All new hires are appointed on a two-year contract in the first instance and will be assessed and co…