Yêu cầu công việc
- Bachelor's degree in Computer Science, Artificial Intelligence, Software Engineering, Data Science or a related field.
- Minimum 3 years' experience building AI systems in production, including at least 1 year on LLM- or agent-based applications used by real users.
- Experience taking responsibility for a component or system through design, development, deployment and operations, with demonstrated expertise in at least one area: LLM and agent systems, AI evaluation and data, or AI platforms and systems integration.
- Strong Python and backend engineering skills, including API development, asynchronous services and databases. Ability to write clean, maintainable code and deploy services using FastAPI and Docker. Working knowledge of CI/CD, monitoring, secure coding, SQL and data pipelines.
- Hands-on experience with LLM APIs and open-weight models, prompt and context engineering, embeddings and vector databases, hybrid retrieval, agent frameworks, tool use and structured outputs. Solid machine learning fundamentals.
- Ability to design evaluation datasets and metrics, use LLM-as-judge, interpret results objectively and use them to guide engineering decisions.
- Experience using AI coding tools and agents effectively in daily engineering work.
- Strong written and verbal communication in Vietnamese and English.
- Alignment with VinUni's core values (EXCEL): Empathy, Creativity, Leadership Mindset, Exceptional Capability, and Entrepreneurial Spirit.
Công việc cụ thể là gì?
ABOUT THE POSITION
The AI Engineer builds and maintains production AI systems for learning, teaching and research at VinUniversity as part of the Center for AI Research's product team. Products include conversational tutors, retrieval over course and academic content, authoring and rubric-based grading assistants, and research agents. The role spans the full stack of a large language model (LLM) application: data, retrieval, agent orchestration, evaluation and operations, using AI tools throughout development. The Product Owner defines product scope and priorities; engineers own technical scoping, implementation, deployment, evaluation and support through user adoption. Senior engineers provide technical leadership, while each engineer leads one or more areas — LLM and agent systems, AI evaluation and data, or AI platforms and systems integration — according to their strengths and product needs.
KEY RESPONSIBILITIES
1. AI Product Development
Requirements & End-to-End Delivery
• Translate business needs and use cases in learning, teaching and research — adaptive and personalized learning, AI-assisted teaching, AI support for academic research — into AI features with clear scope and success criteria.
• Deliver production AI features and backend services integrated with the LMS, single sign-on and University systems; define data collection requirements from the outset.
• Use AI agents and tools across the development lifecycle — research and assessment of emerging technologies, task decomposition, coding, test generation, code review, documentation and monitoring — and continuously improve the team's development practices.
Adoption & Collaboration
• Continue supporting features after release until users have adopted them; translate usability issues, unusual use cases and unmet needs observed in real use into specific, prioritized recommendations for the Product Owner.
• Work with the Product Owner, the AI Product Manager, faculty, researchers and Vingroup's platform and security teams throughout the product lifecycle.
2. AI Systems Engineering
LLM, Retrieval & Agents
• Build production LLM applications with prompt and context management, structured outputs, hybrid retrieval over course and academic content, and agent workflows with tool use, memory and multi-step orchestration (e.g. LangChain, LangGraph, LlamaIndex).
• Improve reliability and cost efficiency by implementing guardrails, fallback mechanisms, human-review steps where AI output requires a person's decision, model routing, caching and per-user usage controls.
Data for AI
• Build the data foundations behind the products: ingestion and chunking of course and academic documents, embeddings and vector stores, access control on learner and research data.
• Prepare datasets for fine-tuning or domain adaptation where justified.
3. Quality, Evaluation & LLMOps
AI Evaluation
• Build and maintain automated evaluations for every release, assessing accuracy, groundedness, hallucinations, safety, guardrail adherence, latency, cost and user value using curated test sets, LLM-as-judge and samples of real usage.
• Build validated reference datasets for AI evaluation with subject-matter experts coordinated by the Product Owner. Validate prompt and model changes against agreed evaluation criteria before adoption.
LLMOps & Operations
• Implement tracing and observability for AI features (e.g. Langfuse); version prompts, models and evaluation sets; run evaluations in CI so regressions are caught before release.
• Monitor production behavior, investigate incidents and quality regressions, and optimize services for quality, latency, reliability and cost.
4. Architecture, Security & Standards
Architecture & Standards
• Contribute to technology roadmaps, technical standards and solution architecture; present and justify technical designs during Vingroup's architecture reviews.
• Generalize solutions into reusable components and patterns; make technical trade-offs explicit and explain them to non-technical colleagues.
AI Application Security & Compliance
• Build AI-specific safeguards into the application: defenses against prompt injection and data leakage, protection of personal data in prompts and logs, and pedagogical guardrails such as tutors that guide rather than give answers.
• Deliver within Vingroup's development, security and AI-risk processes; maintain source code, technical documentation and user guides to the team's standards.
5. Applied Research & Mentoring
Research to Product
• Stay current with advances in LLMs, agents and evaluation; proactively test new tools and research methods against the current system and adopt those that improve results.
• Contribute to controlled experiments on new AI capabilities and document results that can inform product decisions or be reported or published.
Mentoring & Knowledge Sharing
• Mentor and train fresh graduates and junior engineers through hands-on technical guidance, code reviews, pair programming and regular feedback.
• Contribute documentation, demonstrations and internal sessions that help colleagues adopt what you build.
• Ngày làm việc:
• Giờ làm việc:
• Cấp bậc:
• Loại công việc:
