Research Scientist, Voice Program (NLU)

Tokyo, Japan
Rakuten
ago
fulltime NLP

Position Details 

- Work on research projects with senior members to stay on top of the state-of-the-art in NLU, NLG and other areas related to conversational AI and be able to implement and experiment with these models. 

- Work with other research scientists to develop novel algorithms and initiatives that will support the vast amount of business data and resources 

- Join the discussion on short/middle-term direction of projects while aligning with business, product and engineering team  

- Work with Project Management team to gather business requirements, define suitable KPIs and guide members on how to validate if the solution(s) built meets/exceeds the requirements/KPIs  

- Support engineering teams and product teams to build large-scale, integrated speech processing systems, using both in-house developed solutions and external vendor solutions. 

- Contribute to the advancement of science via writing papers, attending conferences and submitting patent proposals   

Mandatory Qualifications: 

- Currently has or is in the process of obtaining a PhD in Computer Science, Machine Learning, NLP or related areas Or MS with 4+ years of industry experience. 

- Strong background and hands-on experience in NLP/NLU/NLG or other areas related to conversational AI. Specifically, experience of working with one or more of these areas –text embeddings, semantic text matching, entity extraction / disambiguation, intent classification, question-answering, chatbots, text summarization, multi-turn dialog management –is expected. 

- Familiarity with open-source ML, DL, NLP software. E.g. Python libraries like Scikit-learn, Tensorflow, Pytorch, and others. 

- Strong programming skills in any programming language, preferably Python.  

- Strong background in data structures and algorithms. 

- Academic publications in top-tier peer reviewed conferences / journals such as ACL, AAAI, IJCAI, EMNLP, NeurIPS, ICML, ICDM, CIKM, KDD or others. 

- English fluency and ability to communicate effectively with researchers, engineers and business people 

- Self-motivated and willingness to take challenges to new technologies, research problems and services 

Desired Qualifications: 

- Practical knowledge of deep learning (DL)-based pretrained neural language models such as BERT, Universal Sentence Encoder or others and experience with pretraining / fine-tuning these models for downstream tasks. 

- Knowledge of training large-scale ML/DL models on CPU/GPU-based hardware. Experience with distributed training 

- Experience with end-to-end product development. 

- Familiarity with DevOps, cloud-related development , and other related development toolkits. 

- Passion for contributing to business by applying innovative solutions to real world problem