AI Engineer

U3 PROJECTS PTE. LTD.; Singapore; On-site; SGD 120,000 - 180,000; Full time

Job summary

U3 PROJECTS PTE. LTD. seeks a platform engineer with a strong AI specialization to build shared platform capabilities for AI applications. The role spans gateway to model providers, workflow runtime, subscription and entitlement handling, and file management systems.

You will design services, integrate model providers, and own deployment/run aspects. This is a platform engineering role focused on enabling other teams to build on top of what you ship.

Qualifications

• Strong backend engineering experience in Python with asynchronous code, type annotations and solid testing practices.
• Experience in codebases relied upon by other engineers daily.
• Proven ability to build services consumed by other teams, not just user-facing apps.
• Hands-on API design, versioning, deprecation, and making breaking changes when needed.
• Experience with multi-tenant platforms: authentication/authorization, quotas, rate limits, usage metering, and cross-tenant isolation.
• Proven delivery of durable workflows/orchestration in production (Temporal preferred, or alternatives with understanding of idempotency, retries, partial failures, and compensation).
• Experience integrating third-party model/API providers at scale (streaming, token accounting, backoffs, graceful degradation, deprecations).
• Hands-on infrastructure capability: Dockerfiles, Helm charts, deployments, Terraform resources, debugging with Kubernetes.
• Positive learning mindset, collaboration, strong analytical and troubleshooting skills.
• Strong written and verbal communication; agile mindset and adaptability.

Responsibilities

• Substantial backend engineering experience in Python, including asynchronous code, type annotations and a genuine testing abit.
• Experience in working in a codebase that other engineers depend on daily.
• Experience in building services consumed by other engineering teams - not only user-facing applications.
• Hands-on experience and able to share and apply concretely on API design, versioning, deprecation and what needed to make a breaking change.
• Experience in multi-tenant platform concerns. Authentication and authorization, quota enforcement, rate limiting, usage metering and cost attribution across tenants who must not affect one another.
• Experience in delivering a durable workflow or orchestration engine in production - Temporal preferred, but Airflow, Step Functions, Cadence or similar is fine if you understand the underlying problems: idempotency, retry semantics, partial failure and compensation.
• Experience in integrating third-party model or API providers at production scale. Streaming responses, token accounting, backoff and retry, graceful degradation, and handling provider deprecations that are not on schedule.
• Hands-on experience in Infrastructure capability, and not just literacy. Able to own service through to running: Docker file, Helm chart, deployment, and the Terraform for the resources it depends on. Able to debug it using Kubernetes.
• Possess positive learning and collaborative mindset.
• Strong analytical, problem-solving and troubleshooting skills.
• Good written and verbal communication skills.
• Agile, fast learner and able to adapt to changes.

Skills

Python backend

Asynchronous programming

Type annotations

Testing

API design

Multi-tenant architecture

Authn/Authz

Deployment orchestration

Docker

Kubernetes

Terraform

Terraform/K8s debugging

CI/CD

Agile / fast learner

Communication

Problem solving

Tools

Temporal

Airflow

Step Functions

Cadence

OpenAI Azure/OpenAI Anthropic APIs

OpenSearch / vector store

Job description

Skillset Requirements

• The successful candidate will build the shared platform capabilities that every AI application at the organization is assembled from - the gateway to model providers, the workflow runtime, subscription and entitlement handling, agent harness features, and file management systems.
• The candidate is a strong software engineer with an AI specialization who is also comfortable in infrastructure - someone who can design a service, integrate model providers properly, and then own how it is deployed and runs.
• This is a platform engineering role in an AI domain, not an applied research role. The work is judged by whether other teams can build on what you ship without having to understand its internals.

Responsibilities:

• Substantial backend engineering experience in Python, including asynchronous code, type annotations and a genuine testing abit.
• Experience in working in a codebase that other engineers depend on daily.
• Experience in building services consumed by other engineering teams - not only user-facing applications.
• Hands-on experience and able to share and apply concretely on API design, versioning, deprecation and what needed to make a breaking change.
• Experience in multi-tenant platform concerns. Authentication and authorization, quota enforcement, rate limiting, usage metering and cost attribution across tenants who must not affect one another.
• Experience in delivering a durable workflow or orchestration engine in production - Temporal preferred, but Airflow, Step Functions, Cadence or similar is fine if you understand the underlying problems: idempotency, retry semantics, partial failure and compensation.
• Experience in integrating third-party model or API providers at production scale. Streaming responses, token accounting, backoff and retry, graceful degradation, and handling provider deprecations that are not on schedule.
• Hands-on experience in Infrastructure capability, and not just literacy. Able to own service through to running: Docker file, Helm chart, deployment, and the Terraform for the resources it depends on. Able to debug it using Kubernetes.
• Possess positive learning and collaborative mindset.
• Strong analytical, problem-solving and troubleshooting skills.
• Good written and verbal communication skills.
• Agile, fast learner and able to adapt to changes.

Good to have:

• Direct experience with Azure OpenAI, OpenAI or Anthropic APIs, including their differing token and streaming semantics.
• Experience in Building evaluation harnesses for LLM-backed systems - regression testing something non-deterministic.
• Experience in TypeScript, for the Next.js and Fastify surfaces that sit in front of services.
• Experience in OpenSearch, or another retrieval or vector store, in a production RAG path.
• Having published and maintained an internal library or SDK, with the version discipline that implies.