Company Name

Honda Research Institute Japan Co., Ltd.

Job Type

Researcher - Research and development of multimodal dialogue technology in human-centered AI

Work Detail

■Recruitment Background Strengthening Our Organization ■Job Description We are seeking a cutting-edge researcher/engineer to advance next-generation dialogue systems. In this position, you will lead research in the intersection of large-scale language models (LLMs) and multimodal signal processing (speech, vision, and text). Your primary mission is to develop a multimodal architecture that deeply understands and adapts to complex human interactions in real time. In collaboration with overseas research centers in Germany and the United States, we will shape the future of human-centric AI technology that transcends language and culture. [Job Details] - Advanced Multimodal Architecture Focus on strengthening internal reasoning (Chain-of-Thought) and reducing latency, and design and verify new research concepts for next-generation dialogue models. - Development of Context Understanding Models Analyze continuous audio, video, and text data to build algorithms that adapt to user behavior and environmental context. - Solve Real-World Problems Design a system that converts real-world interaction signals into high-quality training data, achieving continuous model improvement and bias reduction. - Promote Global Collaboration Work with overseas sister research institutes (Germany and the United States) to tackle research themes with business potential. - Technical Leadership and Growth Responsibilities include creating new research themes, project management, nurturing young researchers, and presenting at international conferences. Expected to play a key role as a team leader in the future. [Job Features] You will be proactively responsible for the entire research lifecycle, from research conception to implementation, evaluation, and practical application. As a bridge between academic research and real-world applications, you will gain experience in transforming fundamental models into operational systems. With a high degree of discretion, you will be able to propose and pursue your own research topic and actively present at top international conferences. - Multimodal Fusion/LLM Based on Vision-Language Model (VLM) and other foundational models, cross-modal attention - Human-centered Computer Vision Speaker detection, gaze estimation, posture estimation, facial expression and attribute analysis - Speech and Natural Language Processing Speaker and speech recognition integrated with visual information, dialogue context understanding Based on Honda's philosophy of "Technology for People," we strive to create unique technologies with speed and creativity. Based on cutting-edge trends in multimodal AI, we will identify high-value research topics and produce results through rapid prototyping, experimentation, and verification. Research results will be actively disseminated through international conferences, papers, etc. You will work in collaboration with team members and several researchers from external research institutions. The appeal of this position is the opportunity to work globally, collaborating with our overseas sister companies (Germany and the United States). Research projects are promoted under the Research Division Manager (Research Division). Research projects are proposed by the researcher himself, including the content and budget, and are launched with board approval. - Approximately half of all employees are foreign nationals, creating a highly international workplace. Communication within the company is conducted daily in both Japanese and English. - The workplace is located within Honda Motor Co.'s Wako Campus, a suburban office building that has won numerous awards. - Researchers have a great deal of discretion, and we actively encourage the external presentation of their research results. We have a culture that respects the company's direction while also valuing the researchers' opinions.

Ideal Profile

Required Skills/Experience [MUST] Master's or PhD in machine learning/AI/computer science (equivalent work experience acceptable) Knowledge of deep learning and multimodal processing across NLP, CV, and speech processing Implementation experience in Python, PyTorch, TensorFlow, etc. 3+ years of work or research experience in a related field Highly collaborative, business-level English proficiency [WANT] Experience with multimodal learning, LLM, and conversational AI Papers presented at top conferences (AAAI, NeurIPS, ICASSP, etc.) Experience using open-source tools (ESPnet, Hugging Face, etc.) Experience training and operating large-scale models Collaborative research experience with overseas research institutions Required Degree Master's or PhD equivalent in computer science, engineering, or a related field If you feel you do not meet all of the above requirements, we still welcome your application. We are looking for candidates who can systematically approach problem-solving and work as a team player to find practical and sustainable solutions in a multicultural environment!

Work Location

Wako City, Saitama Prefecture

Selection Flow

For the selection process, we ask that you submit a resume and a work history in both Japanese and English.

External Site

https://www.jp.honda-ri.com/