QC AI Engineer

COMPANY: JobHub’s Client

POSITION: QC AI engineer

• Level: Senior to Leader

• Experience: 5+ years

• Contract Duration: 2 + years

• Headcount: 03 positions

• Salary range: 55-60M Gross

• Working time: Mon to Fri.

• Employment type: Full-time

• Location: Onsite Gia Lam, Ha Noi

• Client Nationality: Viet Nam

• Interview process: 2-3 rounds

This role is responsible for maintaining the quality and reliability of both AI datasets and Computer Vision / Vision-Language Models (VLMs) prior to deployment in real-world applications. The position contributes directly to building a scalable, high-performing data–model pipeline, supporting AI and robotics systems in production environments.

RESPONSIBILITY

• Oversee and maintain the quality of multimodal datasets (images, videos, text, captions, instructions) used in training and evaluating VLM systems, coordinating with both internal teams and external data providers.

• Establish and standardize data quality guidelines, ensuring consistency across visual content, textual descriptions, and vision-language alignment while identifying noise and bias.

• Conduct data review and auditing processes pre- and post-annotation to detect recurring issues such as incorrect captions, hallucinated outputs, weak prompts, or inconsistent labeling.

• Develop evaluation datasets and testing frameworks for various VLM applications, including visual question answering, instruction following, multimodal reasoning, and interactive AI scenarios.

• Analyze model performance using a combination of quantitative metrics and qualitative assessment, focusing on aspects like accuracy, consistency, robustness, latency, and production stability.

• Perform detailed error analysis and propose actionable improvements related to data quality, prompt design, model retraining, and deployment decisions (go/no-go).

REQUIREMENTS

Background & Experience

• Academic background or practical experience in a relevant field.

• 5+ years of experience in a similar role.

• Strong analytical thinking with a focus on data integrity and model performance.

• Hands-on exposure to AI data workflows, especially in Computer Vision, multimodal systems, or Vision-Language Models (VLMs).

• Good understanding of modern VLM architectures (e.g. vision encoder + language model, alignment techniques, prompting strategies).

• Ability to assess AI outputs beyond traditional metrics, with attention to real-world usability and semantic accuracy.

• Experience handling large datasets and structured evaluation processes.

• Effective communication skills to work with cross-functional teams including engineers, researchers, and external partners.

• Comfortable reading and interpreting technical documents and research papers in English.

Nice to Have

• Experience using data annotation or review platforms (e.g. Label Studio, CVAT, or similar tools).

• Basic proficiency in Python for data analysis or evaluation tasks.

• Familiarity with MLOps concepts and model lifecycle practices.

• Prior collaboration with external vendors or crowdsourced labeling teams.

• Exposure to applications such as robotics, embodied AI, or human–machine interaction.

Personal Attributes

• Proactive mindset with a strong sense of ownership and accountability.

• Detail-oriented with good problem-solving capabilities.

• Eager to learn and adapt in a fast-evolving environment.

• Clear and effective communicator.