Job Description
- Develop computer vision and AI models to transform video into analyzable data, covering object detection, tracking, and recognition tasks.
- Optimize and restructure existing systems based on Python and NVIDIA vision AI models (YOLO, RT-DETR, DeepSORT, etc.) to improve accuracy and efficiency.
- Build and deploy OCR and multi-model workflows (e.g., digital fusion, ROI cropping) to reduce error rates and accelerate inference.
- Research and apply model compression, quantization, and GPU acceleration (CUDA / TensorRT) to enhance real-time processing.
- Collaborate with the AI team to transform structured data into visualization and analytics applications using Generative AI / LLM technologies.
- Develop LLM-based "Game Highlight Generation" modules that combine play-by-play event data with CV features to automatically produce highlight summaries, game recaps, and tactical insights.
- Contribute across the full product cycle: data collection, annotation, model training, evaluation, deployment, and launch.
Qualifications
Experience Requirements
- 4–5+ years of hands-on experience in computer vision / deep learning, with full-cycle project experience (from zero to deployment).
- Proven experience with video processing and Multi-Object Tracking (MOT), including handling fast-moving objects and occlusions.
- Hands-on experience with OCR technologies (Tesseract, EasyOCR, Google Vision API, etc.) and accuracy optimization.
- Experience in NLP or LLM applications (e.g., GPT, LLaMA, Mistral, RAG) to transform structured/unstructured data into summaries, analysis, or natural language output.
Skills Requirements
- Strong proficiency in Python and PyTorch / TensorFlow for model training and inference optimization.
- Solid knowledge of GPU acceleration and deployment (CUDA, TensorRT, ONNX Runtime), including profiling and performance tuning.
- Skilled in video preprocessing, annotation, and dataset management.
- Familiar with multi-model ensemble inference and error-rate optimization strategies.
- Ability to develop multimodal AI systems integrating CV outputs with LLMs for summarization and key insight extraction.
Bonus Points
- Experience deploying on edge devices (Jetson, Coral TPU).
- Background in sports or sports video processing projects.
- Ability to read state-of-the-art CV/AI research papers and quickly implement them in practice.
Language
- Strong Chinese/English reading skills, with the ability to understand technical documentation and write technical reports.
Interview Process
After submitting your resume, please also complete our recruitment questionnaire. Candidates who pass the initial screening will proceed through three stages:
- Initial Interview
(1 hr): With the Co-Founder & CTO. - Technical Assessment
(2–3 hrs): Online coding + model implementation test. - Final Interview
(2 hrs): With Co-Founders & CEO, covering system design, problem-solving, and cultural fit.
All interviews will be conducted online via video conference.
About Preciser
Preciser is a U.S.-based company redefining sports data analytics with AI and computer vision. We primarily serve basketball and baseball organizations worldwide, from schools and leagues to tournaments and professional teams. With our technology, teams and coaches can instantly turn game and practice videos into stats and highlights, and leverage intelligent analytics tools to gain actionable tactical and player insights.
Our mission is to make sports data analytics fast, intelligent, and accessible—empowering stakeholders across the industry to act on insights in real time. We strive to deliver immersive fan engagement, increase sponsorship value, unlock new revenue opportunities, and support teams and organizations in making smarter decisions.
Preciser operates fully remotely, with team members across Taiwan, India, and the U.S. Our team has deep expertise in AI, data analytics, and sports technology. We emphasize practical utility and user experience because we understand that in real games and training, every second and every data point matters.
Since our founding, we have upheld high standards of innovation and product excellence, guided by core values of integrity, teamwork, and customer-centricity. Our vision is to redefine the future of sports through seamlessly integrated data and intelligent analysis—turning data into the ultimate winning edge.
In 2023, we raised an angel round from 500 Global and Silicon Valley angel investors. Within two years, we grew company revenue tenfold while remaining profitable, and expanded our user base beyond the U.S. to Taiwan, Japan, the Philippines, Indonesia, China, and more.