Senior AI Engineer

ABOUT THE ROLE

We are hiring a Senior AI Engineer to own the quality of our live, customer-facing conversational AI agents across both our digital wallet and our new digital bank. This is a hands-on, senior individual-contributor role at the centre of how we take AI agents from idea to production and keep them excellent once live.

You will not be writing the production application code — our engineering squads do that through a structured requirements, sign-off and build process. Instead, you sit on both sides of that process. Upstream, you run the experiments that turn fuzzy ideas into validated, evidence-backed requirements. Downstream, once an agent is live, you own its quality metrics and drive continuous improvement. You are the person who makes sure we build the right thing, and that it stays right in production.

WHAT YOU'LL DO

• Experiment and validate (upstream)
• Design and run experiments to de-risk agent requirements before they reach engineering, spanning model selection, prompt and system-prompt design, retrieval strategies, and classification approaches.
• Turn experiment results into clear, evidence-backed requirements, and co-author product requirement documents alongside the product team so engineering builds the right thing once.
• Bring experimental rigour: frame a hypothesis, choose the right metrics, compare approaches, and make the call based on evidence rather than opinion.
• Own AI quality in production (downstream)
• Own the north-star quality metrics for our live conversational agents across both business lines.
• Continuously observe, run evaluations, and diagnose quality gaps using a structured evaluation framework (golden sets, LLM-as-judge, observability tooling).
• Author and drive improvement plans that span both knowledge and retrieval (RAG) optimisation and agent optimisation (routing, classification, persona, guardrails).
• Partner with engineering to ship structural improvements, and directly tune the AI layer you own (prompts, knowledge content, evaluation sets).
• Scale the craft
• Establish repeatable methods and standards for how we experiment with, evaluate, and operate AI agents, laying the groundwork for a permanent team you may help build and lead.

WHAT YOU'LL BRING

** Must-have

• 4 to 6 years of experience overall, with 2+ years hands-on in production LLM / GenAI systems.
• Strong prompt and context engineering.
• Hands-on experience designing and optimising retrieval-augmented generation (RAG).
• Solid LLM evaluation methodology: defining metrics, building golden and evaluation sets, and LLM-as-judge techniques.
• Proficiency in Python and working directly with LLM and vision-model APIs.
• Genuine experimental rigour and sound model-selection judgement.
• Native or near-native Thai — you must be able to judge the correctness, groundedness, and tone of Thai-language AI output.
• Professional working proficiency in English.

** Strong plus

• Agent orchestration frameworks (e.g. LangGraph or equivalent).
• LLM observability tooling (e.g. LangFuse, LangSmith).
• Evaluation libraries (e.g. DeepEval, Ragas, G-Eval).
• Vector databases (e.g. Pinecone, pgvector).
• Model Context Protocol (MCP) and modern agent tooling.
• Familiarity with agent architecture patterns: routing, classification, guardrails, persona design.
• Experience in fintech, banking, or another regulated, compliance-sensitive environment.

*** Not required for this role

• Training or fine-tuning deep-learning models from scratch.
• MLOps or production-infrastructure engineering.
• Production backend or application engineering — our engineering squads own the build.

WHO YOU ARE

Relentless about quality. You own a number, and you will not let a subtly-wrong or off-tone answer ship. Excellence in production is personal to you.

Persuasive with evidence. You do not own the build, so you win the argument with data, influencing product and engineering through well-run experiments and clear results.

Fluent in both worlds. You translate effortlessly between the business (product, stakeholders) and the technical (engineering), and you are trusted by both.

A proactive owner. You hunt for gaps and regressions before anyone flags them, and you bring a plan, not just a problem.

WHY JOIN

Own AI quality end-to-end for AI agents used by millions of customers.

Work at the frontier of applied GenAI in one of the region's leading fintech companies.

Shape how a whole organisation experiments with, evaluates, and operates AI — and grow into leading a team.