- Company Name
- Overvak
- Job Title
- Computer Vision Engineer (Contract)
- Job Description
-
Job Title: Computer Vision Engineer (Contract)
Role Summary:
Design, implement, and optimize AI pipelines for action detection and classification in real‑time video, integrating models into a camera ecosystem and deploying solutions at scale.
Expactations:
* 3‑month contract with possibility of full‑time conversion.
* Deliver robust end‑to‑end vision models, continuously monitor and improve their performance, and work closely with camera system integration.
* Collaborate with cross‑functional teams, meet deadlines, and contribute to product direction for a global parenting platform.
Key Responsibilities
* Train, fine‑tune, and evaluate PyTorch/TensorFlow models for action detection, classification, and object detection.
* Build preprocessing, data‑augmentation pipelines and real‑time inference workflows.
* Deploy models to cloud (GCP/AWS) and design scalable, edge‑ready inference solutions.
* Implement automated retraining and continuous‑optimization pipelines.
* Research and apply state‑of‑the‑art CV techniques (CNNs, Vision Transformers, efficient detectors).
* Collaborate on camera‑system integration: calibration, lighting, optics, and livestream performance.
* Maintain model monitoring dashboards and troubleshoot live deployments.
Required Skills
* 2+ years CV experience with PyTorch/TensorFlow.
* Deep knowledge of CNNs, Vision Transformers, YOLO, Faster‑R‑CNN, SSD, and related frameworks.
* Python proficiency and libraries: OpenCV, NumPy, scikit‑learn, Pandas, Matplotlib.
* Cloud dev experience (GCP, AWS).
* Real‑time video processing, edge computing, and cloud inference expertise.
* Understanding of camera hardware (calibration, optics, acquisition).
* Experience with VLMs, video search, and summarization agents.
* Optional: Spark, Ray, Kafka, Docker, LangChain, C++.
Required Education & Certifications
* Bachelor’s or Master’s in Computer Science, Engineering, AI/ML, or related field.
* No specific certifications required.