Machine Learning Engineer - Speech AI

In this role, you will join an AI and Robotics team to develop and add new AI-enabled features to a mobile AI communication robot. The team focuses on improving service efficiency for partners using technologies such as Classical and Deep Learning-based Computer Vision, Automatic Speech Recognition (ASR), Text-to-Speech (TTS), and Retrieval Augmented Generation (RAG).

As an expert in Speech AI, you will handle tasks including ASR models, Voice Activity Detection, Language/Emotion Detection, and Speaker Diarization.

Responsibilities

Implement end-to-end speech processing pipelines for various projects. Technical Research: Stay updated on the latest achievements and research in Machine Learning. Scalable Coding: Write performant, high-quality code that can be easily deployed to thousands of mobile robotic units. Cross-Language Integration: Work with Python for general AI codebases and C++ for programs intended to run on edge hardware (Jetson boards).

Requirements

Proficiency in Python and a solid knowledge of C/C++. DevOps Tools: Experience with versio control (Git) and containerization (Docker or Podman). Deep Learning Fundamentals: * Mastery of neural network architectures such as Encoder-Decoder, Transformers, and RNNs. Understanding of core ML concepts including supervised/unsupervised training, classification, and regression.

Familiarity with evaluation metrics like WER/CER and Cross Entropy. Speech AI Expertise: * Knowledge of audio preprocessing, Voice Activity Detection (VAD), and Speaker Diarization. Proficiency with libraries such as the HuggingFace ecosystem, OpenAI Whisper, and NeMo. Nice-to-Have Skills: Education: Master’s degree in Computer Science, Deep Learning, or a related technical field.

Deployment Experience: Practical experience deploying Speech AI systems, including ASR and Emotion Detection. ASR Specifics: Knowledge of model distillation, fine-tuning strategies, and model evaluation. Advanced Systems: Familiarity with distributed systems, cloud computing, and high-performance computing (HPC). Specialized Hardware: Exposure to NVIDIA technologies such as CUDA, TensorRT, or Triton.

Language: Ability to read and write Japanese.

Benefits

Schedule: Flex working schedule (8 hours/day) between 07:00 and 22:00. Remote Work: 2 days per week remote, with potential expansion up to 4 days based on performance. Leave: A long holiday policy allowing up to 1 month of continuous leave. Language: A fully English-speaking work environment within the technology team.

Culture: Monthly or quarterly company-sponsored team meals and recreational events (BBQs, training camps, etc.) Benefit Program: Paid Leave: 15 days of paid leave annually, cumulative for up to 2 years. Allowances: * Housing: 30,000 yen/month Child: 10,000 yen/child/month (up to 2 children, until 14 years of age). Commuter: Full allowance between home and the closest station to the office.

Late-night: Additional allowance for work performed after 10:00 PM. Professional Development: Learning Development Credit Program worth up to 30,000 yen/year for career-related growth. Insurance: Comprehensive Health, Pension, and Employment insurance Family Support: Maternity and paternity leave for up to 1 year (available after 1 year of employment).