6 days ago
Singapore, SingaporeStaff+
Responsibilities
- Design and implement unified end-to-end RL training pipelines using model-based RL, offline RL, simulation, and online learning.
- Create state representations, reward structures, latent world models, and marketplace models for supply-demand interactions and behavioral signals.
- Develop systems that jointly optimize pricing, dispatching, and supply repositioning across multiple marketplace objectives.
- Build policy evaluation and monitoring frameworks for the safe production deployment of decision-making policies.
- Fine-tune open-source LLMs for prediction, reasoning, and simulation in marketplace applications.
- Design language-model training strategies including supervised fine-tuning, preference-based alignment, and iterative improvement.
- Lead a small team technically and guide both research direction and production implementation.
- Collaborate with data engineers and backend engineers to integrate RL and LLM systems into real-time production environments serving millions of users.
Requirements
- PhD in Computer Science, Operations Research, Applied Mathematics, or a related field.
- At least 10 years of experience leading complex technical initiatives across research and production systems.
- Proficiency in reinforcement-learning fundamentals including Markov Decision Processes, stochastic control, and reward-design trade-offs.
- Experience building production ML/RL systems using online learning or simulation-based optimization.
- Knowledge of world models, sequential modeling, latent dynamic models, and representation learning.
- Hands-on production experience fine-tuning LLMs with supervised fine-tuning and at least one of RLHF, DPO, or GRPO.
- Experience with parameter-efficient methods such as LoRA, QLoRA, or PEFT and their accuracy, memory, and cost trade-offs.
- Ability to design generative-model evaluation frameworks covering factual accuracy and reasoning quality.
- Experience with distributed training frameworks such as DeepSpeed, FSDP, or Megatron-LM.
- Proficiency in Python and PyTorch or TensorFlow.
- Experience with scalable computing platforms such as Spark or Ray.
- Ability to translate ambiguous business problems into concrete modeling tasks, data requirements, and success criteria.
Benefits
- Term Life Insurance and comprehensive Medical Insurance.
- GrabFlex benefits package customization.
- Parental leave, Birthday leave, and Love-all-Serve-all volunteering leave.
- Confidential Grabber Assistance Programme.
- FlexWork arrangements including differentiated hours.
- Work from Grab's One-North Singapore office.
Tech Stack
Categories
About Grab
Grab builds a Southeast Asia superapp for mobility, deliveries, and financial services, serving consumers, drivers, merchants, and enterprises. It monetizes via ride and delivery commissions, merchant services (such as ads and subscriptions), and payments and digital banking products. Founded in 2012 and headquartered in Singapore, Grab operates across multiple Southeast Asian markets and is publicly listed on NASDAQ.