Senior Computer Vision Engineer
Vị trí làm việc tại Sigmawave AI, từ xa tại Việt Nam, với hợp đồng full-time. Bạn sẽ chịu trách nhiệm toàn diện về pipeline huấn luyện mô hình phát hiện vật thể, xây dựng quy trình chuẩn chú thích, thử nghiệm, đánh giá, tối ưu và triển khai mô hình AI cho các bài toán thực tế. Mức lương và phúc lợi cạnh tranh tuỳ theo năng lực.
Yêu cầu ứng viên có 5–8 năm kinh nghiệm về computer vision hoặc ML engineering, am hiểu sâu các framework phát hiện vật thể như YOLO, Detectron2, MMDetection. Thành thạo Python, Linux, Docker, các công cụ triển khai như PyTorch, MLflow, DVC, Weights & Biases và môi trường cloud (AWS, GCP, Azure). Ưu tiên kinh nghiệm xây dựng quy trình MLOps và tối ưu mô hình cho production.
About Sigmawave AI
Sigmawave AI leverages world models to bring physical intelligence to real-world environments. We specialize in edge-case intelligent systems — solving the hard, unpredictable scenarios where conventional vision models fall short. Our team builds and deploys production-grade AI that reasons about the physical world, moving fast from research to impact.
The Role
We are looking for a Senior Computer Vision Engineer to own and scale our vision AI training pipeline end-to-end. You will be the person responsible for taking a use case from raw data all the way through to a deployed, policy-governed detection model. This is a hands-on, high-ownership role where you will define how we build, evaluate, and ship vision models.
What you will do
- Design and build the training pipeline: Architect a robust, reproducible pipeline for training and fine-tuning object detection models (YOLO, Detectron2, MMDetection, or similar) from data ingestion through to model registry.
- Own the data strategy: Define annotation standards, manage labeling workflows (e.g., CVAT, Label Studio, Roboflow), and build data versioning and augmentation pipelines.
- Define deployment policies per use case: Translate business requirements into technical policy plans covering confidence thresholds, inference mode (edge vs. cloud), model update cadence, fallback logic, and alerting rules.
- Evaluate and benchmark models: Establish evaluation frameworks (mAP, precision/recall, latency) and build dashboards to track model performance across use cases.
- Deploy and monitor: Package models for production (ONNX, TensorRT, Triton) and set up monitoring for data drift, model degradation, and real-time performance.
- Collaborate cross-functionally: Work closely with product, operations, and client-facing teams to scope use cases, set expectations, and document policy plans for each deployment.
What we are looking for
- 5–8 years of experience in computer vision, machine learning engineering, or a closely related field.
- Deep hands-on expertise with object detection frameworks such as YOLOv5/v7/v8, Detectron2, or MMDetection.
- Strong understanding of model training lifecycle: data collection, annotation, augmentation, training, hyperparameter tuning, evaluation, and versioning.
- Experience building and maintaining ML training pipelines using tools like PyTorch, MLflow, Weights & Biases, DVC, or equivalent.
- Proven ability to define and document deployment policies, including threshold tuning, retraining triggers, and use-case-specific inference strategies.
- Familiarity with model optimization for production: ONNX, TensorRT, quantization, edge deployment.
- Solid Python skills and comfort with Linux, Docker, and cloud environments (AWS, GCP, or Azure).
- Strong communicator who can translate technical decisions into clear policy documentation for non-technical stakeholders.
Nice to have
- Experience with video analytics, multi-camera tracking, or real-time streaming pipelines.
- Background in segmentation, pose estimation, or OCR in addition to detection.
- Exposure to synthetic data generation or active learning workflows.
- Prior experience in a startup or fast-paced applied AI environment.
- Contributions to open-source CV projects or published research.
Our Stack
- Frameworks: PyTorch, Ultralytics YOLO, Detectron2, OpenCV
- MLOps: MLflow, Weights & Biases, DVC, Docker
- Inference: ONNX Runtime, TensorRT, Triton Inference Server
- Annotation: CVAT, Label Studio, Roboflow
- Cloud: AWS (SageMaker, S3, EC2) / GCP
- Languages: Python, Bash, SQL
Why Sigmawave AI
- High-impact work: your models will be deployed in real-world environments solving tangible problems.
- Ownership culture: you will have end-to-end ownership of the vision pipeline, not just a piece of it.
- Cutting-edge stack: we invest in the best tools and infrastructure so you can move fast.
- Growth: work alongside a team of sharp engineers and researchers pushing the boundaries of applied AI.
How to apply
Send your resume and a brief note on a CV project you are proud of to karchun@sigmawave.ai and info@sigmawave.ai. We review applications on a rolling basis.
Similar Jobs
