AI Engineer II, Voicebot

MoMo is Vietnam’s leading mobile-payments platform, on a mission to make every transaction fast, easy, and joyful. In our AI CreditTech team, Machine Learning is the engine behind the products millions of Vietnamese use every day: real-time credit scoring and loan decisioning, transaction verification, and AI-driven collections. We are looking for a Machine Learning Engineer who can own ML features end-to-end, from problem framing and modeling to shipping low-latency services in production and keeping them healthy. Mô tả công việcDesign, build, and own ML solutions for fintech problems: credit scoring, decision and routing engines, fraud and transaction verification, recommendation, user segmentation, and recovery automation. Take models from prototype to production, building and productionizing training and inference pipelines that serve real-time traffic at scale (sub-200ms latency, high availability). Partner with Data Engineers, Analysts, Risk, and Product to turn business problems into measurable ML outcomes. Run rigorous experimentation (A/B tests, backtesting, and simulation) and use the results to drive model and product decisions. Monitor deployed models for drift, performance, and data-quality issues, owning diagnosis and iteration when metrics move. Contribute to our shared ML platform, tooling, and MLOps practices, and help raise the engineering bar through code reviews and design discussions. Explore applied GenAI and agentic systems (LLM-based assistants, retrieval, and workflow automation) where they add real value. Yêu cầu công việc2–4 years of hands-on experience building and shipping ML systems to production (fintech, large-scale consumer, or real-time systems a strong plus). Strong programming skills in Python (production-grade, not just notebooks); working knowledge of Java is a plus. Practical experience with the modern ML/data stack: Scikit-learn (and/or PyTorch/TensorFlow), FastAPI for model serving, Airflow for orchestration, and Kafka / Spark / Lakehouse or BigQuery for data. Solid grounding in probability, statistics, and algorithms, and sound judgment about model evaluation, validation, and trade-offs. Comfortable owning a feature end-to-end and collaborating across Data, Risk, and Product, you communicate clearly and reason about business impact, not just model metrics. Exposure to MLOps (monitoring, CI/CD for ML), experimentation frameworks, or applied LLM/agentic tooling (e.g., LangChain, vector databases) is a strong plus.