What you will own

• Build AI-powered features, pipelines, prototypes, and integrations.
• Evaluate accuracy, latency, cost, reliability, and failure modes.
• Work with product and engineering teams to define practical AI use cases.
• Prepare datasets, prompts, experiments, and evaluation methods where needed.
• Integrate AI services or models into production-oriented applications.
• Document assumptions, limitations, and operating requirements.

 

Core technology stack

• AI/ML engineering: Python, AI/ML or LLM APIs, prompt/workflow design, evaluation datasets, embeddings/retrieval, and experiment tracking practices.
• Application integration: Backend APIs, service integrations, queues or background jobs, data pipelines, and product-facing AI workflows.
• Data/evaluation: Structured and unstructured data processing, test sets, metrics, failure analysis, and reproducible evaluation harnesses.
• Delivery: Git, Docker/containerized environments, cloud deployment patterns, monitoring, logging, and secure handling of model/data access.

What we are looking for

• Practical experience building AI/ML or LLM-based workflows.
• Ability to evaluate model/system behavior with evidence.
• Solid software engineering ability, especially Python/API integration.
• Understanding of failure modes, cost, latency, privacy, and data quality.
• Ability to communicate uncertainty and limitations clearly, including when a simpler non-AI approach may be more appropriate.
• Willingness to travel up to 50% of the time, on short notice when required (see Travel).
• Professional working English.

 

Travel

This role requires a willingness to travel up to 50% of the time. Travel is need-driven rather than scheduled — when a client engagement, on-site deployment, or operational requirement arises, you are expected to travel as required, including across Asia Pacific and the Middle East.

 

Nice to have

• Experience with LLM applications, retrieval, evaluation, or prompt/workflow design.
• Experience integrating AI into web products or backend systems.
• Experience with monitoring, fallback behavior, or human-in-the-loop workflows.

 

First 90 days

• Understand the relevant systems, workflow, delivery context, and current risks.
• Deliver scoped improvements with clear communication and review.
• Reduce bottleneck pressure in your area by leaving behind maintainable code, documentation, checks, runbooks, or reusable patterns.

 

How we evaluate

We evaluate candidates through baseline design, evaluation approach, failure-mode analysis, API/integration boundaries, cost/latency/privacy judgment, and practical awareness of when a simpler non-AI solution may be better.