Lead Machine Learning Engineer (Foundation Models)

GrabTaxi Holdings Pte. Ltd.; Singapore; On-site; SGD 180,000 - 280,000; Full time

Benefits offered by this job

Term Life Insurance

Medical Insurance

FlexWork hours

Job summary

GrabTaxi Holdings Pte. Ltd. is seeking an experienced Lead Machine Learning Engineer to own end-to-end foundation model strategies for the Grab marketplace.

The role is onsite at our Singapore One-North headquarters and reports to the Senior ML Engineering Manager. You will drive pre-training, scalable distributed training, and integration of models into live production systems powering millions of users daily.

Qualifications

• 8+ years of professional ML experience with deep learning and transformers.
• Hands-on pre-training of foundation models (e.g., Llama, Qwen) from scratch.
• Strong background in distributed training across multi-node GPUs.
• Proficient Python/C++ and MLOps deployment.

Responsibilities

• Architect and optimize pre-training pipelines for large foundation models.
• Build and operate scalable distributed training across multi-node clusters.
• Design generative recommendation systems and integrate with production.
• Bridge research and production with cross-functional teams.
• Mentor engineers and enforce high-quality code and testing.

Skills

Deep learning

Transformer architectures

Distributed systems

Python

C++

MLOps

Pre-training foundation models

LoRA/QLoRA

Tools

PyTorch FSDP

DeepSpeed

Megatron-LM

Ray

Job description

Get to Know the Role

This is an applied research and machine learning engineering role aimed at developing foundation model solutions for Grab's massive marketplace. As a Lead Machine Learning Engineer (based onsite in our Singapore One-North headquarters, reporting to the Senior Machine Learning Engineering Manager), you will own the end-to-end lifecycle of our proprietary foundation models and generative recommenders. You will bridge the gap between modern ML research and optimised, large-scale distributed training infrastructure. You will build architectures that unify search, retrieval, and ranking to power decisions for millions of users daily.

The Critical Tasks You Will Perform

1. Architect Pre-training Pipelines: Design, implement, and orchestrate efficient, scalable pre-training pipelines for large language and multimodal foundation models, from raw data curation and tokenization through to converged checkpoints.
2. Build Distributed Training Systems: Develop and operate large-scale distributed training frameworks across multi-node GPU clusters, applying advanced parallelism strategies (FSDP, DeepSpeed, Megatron-LM, Tensor/Pipeline/Expert parallelism) and memory-efficient training techniques.
3. Pioneer Generative Recommendation: Design and deploy generative recommendation systems that combine foundation model pre-training with a full post-training loop (SFT, distillation, preference alignment and RL) to unify retrieval and ranking against marketplace objectives.
4. Optimize Model Architectures: Design, customise, and implement highly efficient transformer architectures tailored for Grab's business use cases, including Mixture-of-Experts (MoE), long-context attention, and tokenisation strategies.
5. Enforce Engineering Excellence: Lead code and design reviews, establish high-standard engineering patterns, and mentor junior engineers to sustain, testable, and production‑grade codebases.
6. Bridge Research and Production: Partner with cross‑functional product, platform, and infrastructure teams to integrate custom foundation models and generative recommenders into live, highly‑available production environments.

What Essential Skills You Will Need

• Industry Experience: At least 8 years of professional experience in machine learning, with a deep focus on deep learning, transformer‑based architectures, and the software development lifecycle.
• Pre‑training Expertise: Proven hands‑on track record of pre‑training or continually pre‑training open foundation models (e.g., Llama, Qwen, DeepSeek, Mistral) from scratch, rather than solely utilising third‑party APIs.
• Distributed Systems Mastery: Deep technical proficiency with multi‑node, multi‑GPU scaling frameworks such as PyTorch FSDP, DeepSpeed, Megatron‑LM, and Ray, along with an understanding of modern hardware accelerators.
• Post‑Training & Alignment: 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).
• Modern Software Engineering: High proficiency in writing clean, maintainable, and testable production code in Python/C++, with solid experience in MLOps/LLMOps deployment pipelines.
• AI & Productivity Adoption: You have experience engaging with AI tools and emerging technologies (such as LLM assistants, code generators, and specialised developer agents) to enhance personal productivity, optimise engineering workflows, and contribute innovative platform ideas.
• Generative Recommendation: Exposure to generative retrieval, semantic tokenisation, sequence modelling of user behaviour, or unifying retrieval and ranking with preference‑aligned models is a plus.

Life at Grab

We care about your well‑being at Grab, here are some of the global benefits we offer:

• We have your back with Term Life Insurance and comprehensive Medical Insurance.
• With GrabFlex, create a benefits package that suits your needs and aspirations.
• Celebrate moments that matter in life with loved ones through Parental and Birthday leave, and give back to your communities through Love-all-Serve-all (LASA) volunteering leave.
• We have a confidential Grabber Assistance Programme to guide and uplift you and your loved ones through life's challenges.
• Balancing personal commitments and life's demands are made easier with our FlexWork arrangements such as differentiated hours.

What We Stand For at Grab

We are committed to building an inclusive and equitable workplace that enables diverse Grabbers to grow and perform at their best. As an equal opportunity employer, we consider all candidates fairly and equally regardless of nationality, ethnicity, religion, age, gender identity, sexual orientation, family commitments, physical and mental impairments or disabilities, and other attributes that make them unique.