[What the role is]

[What you will be working on]

Singapore's land transport network moves millions of people every day. As AI reshapes what's possible in transport operations, the Land Transport Authority is investing in AI capabilities that will meaningfully improve how the network is sensed, managed, and optimised — for commuters, operators, and public officers alike.

You will be part of LTA's Transport AI Programme team, building and maintaining AI-powered applications and services that improve how Singapore's land transport network is managed and experienced. Working closely with data scientists, AI architects, and operations teams, you will translate AI models and capabilities into reliable, production-grade software that public officers and commuters can depend on.

This role reports to the Transport AI Programme Deputy Director and works in close collaboration with industry partners on joint development efforts.

• Design, develop, and maintain AI-powered applications across transport use cases including public transport operations, traffic management, and road operations.

• Integrate large language models, computer vision models, and other AI/ML components into operational workflows, including retrieval-augmented generation (RAG) pipelines and agentic systems.

• Model Optimization & Quantization: Apply and benchmark advanced techniques (e.g., quantization, pruning, distillation, speculative decoding) to optimize LLMs and other deep learning models for inference.

• Evaluate inference engines (vLLM, TensorRT-LLM, Triton Inference Server) and publish recommendations balancing performance, cost, and resilience. Provide hands-on support to agency teams applying these to their workloads.

• Build and maintain APIs, microservices, and data pipelines that connect AI capabilities to operational systems.

• Implement guardrails, evaluation frameworks, and observability tooling to ensure AI systems behave safely and reliably in production.

• Collaborate with industry partners on co-development efforts, maintaining code quality, documentation standards, and integration discipline across organisational boundaries.

• Participate in red-teaming exercises, prompt-injection testing, and other AI safety activities.

• Support deployment, monitoring, and continuous improvement of AI products in production environments.

[What we are looking for]

• Knowledge in Computer Science, Computer Engineering or a related field and will have 3–6 years of software development experience, with at least 2 years working on AI or ML-integrated systems. Candidates will be assessed based on their experience and may be considered at the intermediate or senior level.

• Candidate should be proficient in Python and familiar with modern AI/ML frameworks such as LangChain, LlamaIndex, PyTorch, or equivalent.

• Work on areas such as LLM integration, agentic workflows, evaluation harnesses, observability, guardrails, model access, retrieval/memory patterns, and multimodal AI use cases.

• Use AI coding and development tools to accelerate delivery while retaining human accountability for design and quality.

• Experience with cloud-native development (AWS, Azure, or GCP), containerisation (Docker/Kubernetes), and CI/CD pipelines is expected. Familiarity with responsible AI practices — such as evaluation harnesses, hallucination mitigation, and model monitoring — is a strong advantage.

• Experience working in regulated or government environments is a plus.

• Comfortable working in cross-functional teams and collaborating with both technical and non-technical stakeholders to translate operational data needs into robust engineering solutions.

• Strong communication skills to explain technical concepts to both engineers and non-technical stakeholders. Inclination to work in a collaborative, fast-moving Agile environment. Strong in communication with the ability to explain complex technical concepts clearly to diverse audience.

As part of the shortlisting process for the role, you may be required to complete a medical declaration and / or undergo further assessment.