【Speech AI】Machine Learning Engineer

About avatarin Inc.avatarin was established as a spin-off of ANA Holdings Inc. in 2020 to democratize mobility by opening the door to a world in which anyone can instantly and sustainably transport themselves to a remote destination. avatarin aims to achieve this goal through the development of core technologies that enable real-time teleportation of human presence and skills through robots and other mobility solutions. This new avatarin capability will not only help the world efficiently share its skills but also exponentially expand the spectrum of human interaction data that can be collected from real-world experiences.avatarin’s flagship product is a mobile, communication AI avatar robot called newme. Deployment of the newme robot is the first step in a larger vision to pioneer an instant, sustainable, and inclusive mobility network to connect people to places and experiences. avatarin has been operating newme robots in aquariums and museums since 2021 and is working to expand deployment to public spaces including airports, hotels, hospitals, government offices, train stations, and retail stores in Japan and around the world.■Job DescriptionAs an AI Engineer in avatarin “AI and Robotics” team, you will work on adding new AI-enabled features to our flagship robot Newme.The AI and Robotic team is working on solutions to improve the service efficiency of our business partners. Projects involve technologies such as Classical and Deep Learning based Computer Vision, Automatic Speech Recognition, Text to Speech, Retrieval Augmented Generation and so on.As an expert on Speech AI, you will be working on tasks involving Automatic Speech Recognition models, Voice Activity Detection, Language Detection, Emotion Detection, Speaker Diarization, Audio CleaningAI related codebases are mainly written in python, but programs meant to run on Newme (Jetson board) are written in C++ for ease of integration. You should be able to work with both language.■ResponsibilitiesImplement speech processing pipelines for our customer projectsResearch on the latest achievements in Machine learningWrite performant code that can be easily deployed to thousands of Newme■Must-have skills◉Programming proficiency in Python and solid knowledge of C/C++◉Version control (i.e Git) and containerization (i.e Docker or Podman)◉Deep learning fundamentals for language processing-Neural network architecture (Encoder - Decoder, Transformers, RNN)-Core ML concept (supervised/unsupervised training, classification, regression, …)-Evaluation metrics (WER/CER, Cross Entropy, …)◉Speech AI Fundamentals-Audio preprocessing-Voice Activity Detection-Speaker Diarization◉Speech AI libraries-HuggingFace ecosystem-OpenAI Whisper-Nemo■Nice to have◉Master’s degree in Computer Science or Deep Learning related field◉Practical experience with deploying Speech AI systems -Automatic Speech Recognition systems -Speaker Diarization -Audio Emotion Detection◉Knowledge on ASR specifics -Model distillation -Model Evaluation -Fine tuning strategies◉Good knowledge on distributed systems, cloud and high-performance computing.◉Good software engineering fundamentals including system design, testing and debugging.◉Familiarity with Nvidia technologies (Cuda/TensorRT/Triton)◉Ability to Read/Write Japanese■Team cultureStrong collaborating and communicating skills, with a passion for learning.Being a valued team player in a dynamic, autonomous, cross-functional team.Love working on a fast-paced team that is constantly learning, experimenting, and iterating.Having a passion for performance excellence and robustness with an engineering mindset.Having an enthusiastic, go-getter attitude.■Selection ProcessesPersonality test（25min)Skills Screening Test (90min)1st interview with HR Manager (45min)Deep Dive Technical Interview (90min)Final Interview (60min) Working ConditionsFlex working schedule 8 hours/day, 5 days/week (Between 07:00 to 22:00)2 days/week remote work (Expansion up to 4 days based on performance)Long holiday policy (up to 1 month continuously)Monthly/Quarterly company sponsored Team Lunch/Dinner eventsCompany wide recreational events (arranged lunches, BBQs, training camps, etc.)Fully English-speaking work environment (Technology team)【avatarin benefit program】15 days paid leave yearly (cumulative up to 2 years)Commuter allowance between company and closest home station.Housing allowance of 30,000yen/month (within a 5 km radius from company).Child allowance of 10,000yen/child/month (up to 2 children & 14 years of age)Learning Development Credit Program up to 30,000 yen/year (must be career growth-related)Late-night work allowance after 10pm.Health, Pension, Employment insurance (50:50 Employee:Company commitment)Maternity/Paternity leave (Up to 1 year, after 1 year employment)HP : https://about.avatarin.com/en/https://www.tokyoupdates.metro.tokyo.lg.jp/en/post-1079/