FPT Smart Cloud is hiring: AI Research (Computer Vision) (Hanoi)
FPT Smart Cloud is hiring a computer vision and multimodal AI researcher in Hanoi for products such as eKYC, OCR, document processing and its Vision Agent Platform. 2+ years of deep learning with Python and PyTorch or TensorFlow required.
The role
At FPT Smart Cloud (FCI), this researcher works on the computer vision and multimodal AI behind products such as the Vision Agent Platform, eKYC, Intelligent Document Processing (IDP) and OCR. The deep learning models span many tasks: detection, tracking, segmentation, OCR, document understanding, face recognition, face anti-spoofing, video understanding and Vision-Language Models (VLMs). A big part of the job is taking new methods from top AI conferences and open-source projects, getting them to run and then making them better; you design the experiments and judge results with suitable metrics. Then you build training pipelines, make models fast for deployment (TensorRT, ONNX) and help software engineers put them into production.
What they ask for
- 2+ years in computer vision, deep learning or AI research; strong Python with PyTorch or TensorFlow
- A firm grasp of modern deep learning (CNNs, transformers) and of the vision tasks above
- Has trained and tuned models on large datasets, designs experiments and improves models from evaluation results
- Can read, reproduce and implement ideas from research papers
- Comfortable with Git, Linux and Docker
Nice to have
- A master's or PhD in computer science, AI, machine learning, computer vision or a related field
- Papers at top venues such as CVPR, ICCV, ECCV, NeurIPS, ICML, ICLR or AAAI
- Experience with VLMs, deploying models with TensorRT, ONNX Runtime, Triton or DeepStream, or running AI on NVIDIA Jetson
Good to know
- Location: Hanoi, in Vietnam: 22nd floor of Vinacomin Tower, 3 Duong Dinh Nghe, Cau Giay; office hours 8:30 to 17:30 with a lunch break from 12:00 to 13:00.
- FPT Smart Cloud's website dates the job page, posted or last updated, 12 November 2026.
How the interview usually goes
Interviews for computer vision research roles usually centre on deep learning fundamentals (CNNs, transformers, loss functions, evaluation) and the field's models and papers, such as detection, OCR or VLMs. Be ready to walk through an experiment or project in detail; there may be Python or PyTorch coding, and an ML system design question on getting a model into production fast and lean.
