MoMo is the market leader in mobile payments in Vietnam. AI is now embedded in nearly every surface of the product — recommendation and personalization, conversational assistants, eKYC, risk and fraud scoring, promotion targeting and financial services. We are looking for an AI Engineer I / AI Engineer II to join Moni, a product squad under the AI Products department (AI Data). Moni ships end-to-end: AI, mobile, backend and QC engineers sit in the same squad, so the features you build go straight to real users — and you see how they behave there. You will build LLM- and agent-powered features against real production traffic, starting from well-scoped pieces with close guidance from senior engineers and growing into owning whole features and making sound engineering calls on your own. Mô tả công việcImplement and ship LLM- and agent-based features — prompts and context, tool calls, service code, tests, rollout — with design guidance from senior engineers (level I) or owning the feature end to end (level II). Build and maintain RAG pipelines: ingestion, chunking, indexing, retrieval and reranking; measure retrieval quality and improve it with evidence. Build agent workflows — tool/function calling, multi-step execution, memory and context handling — and integrate them with internal MoMo services. Write and maintain the evaluation for what you build: eval sets, automated judges, regression runs on every prompt or model change. Report what moved and why; nothing ships on vibes. Version, test and review prompts and context like code, not like text pasted into a config. Work with real data: build datasets, write annotation guidelines, and do error analysis on production traffic and conversation logs. Support the production life of what you ship: tracing, token/cost tracking, dashboards, alert triage, and fixing what breaks. Apply safety and compliance practice by default — prompt-injection resistance, PII handling, guardrails — as a regulated financial product requires. Grow deliberately: reproduce relevant work, bring back evaluated findings on new models and techniques, and treat code review as the main way you level up. Yêu cầu công việcMust haveBachelor's degree in Computer Science, Mathematics, AI or a related field, or an equivalent practical background.0–4 years of professional experience: AI Engineer I — 0–2 years, no production AI track record required; AI Engineer II — 2–4 years, including at least 1 year shipping AI/ML or LLM features to production users. Strong Python: clean, readable, tested code — not just scripts and notebooks. Solid fundamentals — how transformers, tokenization, embeddings and context windows actually work, and the practical failure modes of LLMs (hallucination, drift, prompt injection). Hands-on work with LLM APIs and the modern toolchain — foundation-model APIs (Claude, GPT, Gemini) and/or open-weight models (Llama, Qwen), plus a RAG or agent framework. At level I this can be an internship, side project or competition, as long as you can explain the design decisions behind it; at level II we expect at least one feature you took from prototype to live traffic. Evidence that you can tell whether your own work is good: at level I, a clear way of checking your results; at level II, an evaluation you designed for your own feature and can defend in review. Git, SQL, basic services/APIs, comfortable working in Linux. At level II, also deployment basics — containers, CI/CD, observability, running inference on GCP/AWS. Working English for technical reading and written communication. Learns fast, takes review feedback well, and speaks up early when stuck. Nice to haveFine-tuning experience (SFT, LoRA/QLoRA) with measured gains. Multimodal (vision-language, speech) or small / on-device model work. Vietnamese-language NLP, or experience with fintech and regulated data. Open-source contributions, competition results, or public technical writing.