Tech Lead Engineer (Voice/Speech)

About this role

This role serves as the technical lead for JAPAN AI SPEECH, a product that transcribes meetings and business discussions in real time, separates speakers, and structures decisions and ToDos. The position owns technical decisions for the speech product and works with the Product Manager and engineering teams.

Responsibilities include defining the technical strategy and architecture for the full audio-processing pipeline, from streaming audio ingestion through recognition, speaker separation, and summarization. The role balances accuracy, latency, and GPU inference cost while accounting for customer-specific terminology, translation across more than 60 languages, and processing restricted to the Japan region. It also covers technology selection, technical-debt planning, scalability, and cost design.

The Tech Lead leads the design and implementation of product capabilities such as cross-meeting analysis, LLM-based summarization and issue extraction, audio ingestion from Zoom, Teams, and Meet, and integrations with business tools including Salesforce and Slack. Enterprise requirements such as ISMS and P-Mark compliance, data encryption, and Japan-region processing must be incorporated into the system design. The role is expected to contribute directly to design and implementation, conduct design and code reviews, mentor team members, improve development processes, and participate in technical interviews.

Required qualifications include a bachelor’s degree in computer science, software engineering, artificial intelligence, machine learning, mathematics, physics, or a related field, or equivalent professional experience. Candidates need at least five years of professional software engineering experience and at least three years of experience as a technical lead or architect, including technology selection, design decisions, and technical guidance for members. Experience designing and operating large-scale backend systems with high availability, scalability, and cost considerations is required. Practical experience with cloud services such as GCP, AWS, or Azure and with Docker and Kubernetes is also required. Candidates must have experience with either real-time streaming processing, including technologies such as WebSocket, gRPC, Kafka, or Pub/Sub, or production design and operation of machine-learning model inference infrastructure. Experience in a startup or growth-stage environment, or an equivalent environment, is also listed as required.

Preferred experience includes speech recognition, speaker separation, speech signal processing, GPU inference optimization using technologies such as Triton, TorchServe, or ONNX Runtime, batching or quantization, LLM product development, MLOps pipelines, large-scale data platforms such as BigQuery or data lakes, security-constrained system design, telephone or CTI integrations, new-product launches from the 0-to-1 stage, and full-stack development with TypeScript and React. Japanese at business level or fluent English is required; Japanese-language interviews and work are expected for the business-level Japanese option. Japanese-English bilingual ability is welcomed.

The development environment includes Python for the backend; TypeScript, React, Next.js, and NX for the frontend; LangChain, LangGraph, and the JAPAN AI STUDIO SDK for AI and LLM work; GCP container and Kubernetes infrastructure; Docker; BigQuery; PostgreSQL; and customer data sources. Collaboration and development tools include Slack, Confluence, Linear, Google Workspace, GitHub, Notion AI, Claude Code MAX Plan, Cursor, ChatGPT, and Devin. The work environment provides a Mac with Apple Silicon and supports dual monitors.

The Speech team has five members and works with the Product Manager, the AI Platform team responsible for GPU and Kubernetes infrastructure, and Product Engineers. The engineering organization has approximately 60 members, and the company states that its total workforce is 256 people.

The position is full-time and based at the Shinjuku, Tokyo office, with hybrid work consisting of three days in the office and two remote days per week. Working hours are 10:00–19:00, with Saturdays, Sundays, and public holidays off. Annual salary is ¥7,000,000–¥16,000,000; the posting also lists monthly pay of ¥500,000–¥1,142,857, fixed overtime allowances for 45 hours, twice-yearly salary review opportunities, twice-yearly bonuses, and a stock-option program. Salary may be discussed based on experience, ability, and previous compensation. The probationary period is one month.

Benefits listed include social insurance, full transportation reimbursement, book-purchase assistance, refresh allowances, club allowances, housing assistance in designated areas, quarterly lunch and dinner support, qualification and language-learning support, refresh leave after three years of continued employment, annual health checks, and an employee shareholding plan. The selection process is document screening, a coding test, f