Design, develop, and operate a data distribution system for various products using advanced AI technology.
Machine Learning Engineer (Computer Vision and Technical Drawing Structuring)
About this role
This Machine Learning Engineer role develops general-purpose algorithms that convert plant design drawings and mechanical drawings into data that can be processed, searched, and used in solutions. The work includes designing and implementing algorithms that recognize symbols and extract dimensions and annotations from drawing images and CAD data.
The engineer will build models and preprocessing methods that remain robust against drawing-specific noise, including handwritten annotations, degraded scans, and varied drafting conventions. Individual issues identified by FDEs in customer projects are examined through longer-term research and development and transformed into generalized algorithms. The role also includes preparing evaluation metrics and benchmark datasets for drawing structuring, and following research in areas such as image recognition and graph-structure extraction for application to internal technology.
The position is part of the PaaS Unit. The unit develops SAT (Stockmark A Technology), a platform for implementing business AI, and works in a broader organization that develops its own Japanese-document-focused LLM and VLM technologies. The posting specifically names Stockmark-2-VL-100B and describes applying and extending the company’s VLM technology to the difficult domain of technical drawings. The role connects customer-project findings from FDEs with reusable research and product technology.
The listed development environment includes Python, TypeScript, Vue.js, and Node.js. The container technology is Docker, infrastructure as code uses Terraform, and the listed cloud platforms are AWS and Azure.
The PaaS Unit consists of one business lead who is also the CEO, one development lead, seven product engineers, five RAG engineers, three machine learning engineers, one product manager, one researcher, one customer success member, three business members, and five new-business planning members.
The employment location is the Tokyo headquarters or the employee’s home or another place where remote work is possible. Full remote work is allowed and is described as the basic working arrangement. Residence must be within Japan; full remote work while residing overseas is not allowed. Residents within the company’s commuting area of Tokyo and neighboring prefectures may freely choose between remote work and office attendance without a frequency limit. Residents outside that area may voluntarily attend the headquarters once per month. The scope of duties may change to development and operation of the company’s services and products, related work, and other duties designated by the company.
This is an AI-generated summary of the employer's original posting — details can be incomplete, out of date or simply wrong. Always confirm everything on the official posting before applying.