バックエンドエンジニア

必須

• Currently residing in Japan, or willing and able to relocate to Japan
• Business-level Japanese communication skills (equivalent to JLPT N1 or higher)
• Backend development experience for web or smartphone applications
• Experience designing backend APIs and databases
• Experience with agile development at an operating company (事業会社)
• Experience conducting code reviews and design reviews

歓迎

• Experience leading technology selection and architecture design
• Experience processing data with generative AI
• Experience working with stakeholders to define and agree on requirements
• Experience at a startup or in 0→1 product development
• Experience building and operating data platforms or cloud infrastructure

求める人物像

• You share SalesNow's mission and our three values (AI Native, Kotoshikou — being driven by purpose, not by tasks — and Highest Quality), and you are determined to create value by making full use of AI
• You enjoy moving fast. Even as things change, you start from the purpose, find the problem, and see it through to a solution
• You work with your team and customers with integrity and ownership
• You absorb new technologies and knowledge, and take on product development that drives the business forward
• You take the voices of customers and business teams into account, and solve fundamental problems through the product
• You check logs and data before deciding, instead of trusting assumptions

応募について

• ※This position is also open to non-Japanese candidates who are able to work in Japan as full-time employees.

使用ツール

Why We Are Hiring

• At the center of the product is our Python (FastAPI) product API. 780 pull requests were merged into it in the last 30 days (measured on 2026-09-12). You will help design how the team verifies changes at this pace. SalesNow is approaching Series B, and you can shape the backend architecture and engineering practices as the product enters its next growth phase.
• Working with AI has increased how much code we can ship. Engineers exercise judgment by checking whether each change fits the architecture and handles the underlying data correctly. We are looking for a backend engineer who can improve search performance, redesign the model layer, and improve the quality of LLM features.

Problems You Will Work On

• As of 2026-09-12, these are open:
• Reduce full reindexing time for our OpenSearch platform, which holds 105 million documents (about 460 GB) and supports searches across companies, offices, and departments. Source-side DB queries and the ingestion process are the candidates for improvement, and you will redesign them
• Break up a 2,800-line model layer without stopping the service. A four-phase plan is in place to resolve circular imports and clarify layer boundaries, and you will refine and execute it. It uses the Strangler Fig pattern, feature flags, and API versioning to preserve backward compatibility. This work runs alongside incident response

Responsibilities

• You will be responsible for backend development at the core of SalesNow. The scope will be adjusted to your experience and interests.
• Backend development of generative AI and data features using Python (FastAPI) and TypeScript
• Design, PoC, and production implementation of new LLM-based features
• Structuring, summarizing, and interpreting company data with LLMs
• Using LLMs to prepare company information and suggested next steps for salespeople to review and act on
• Fast product iteration with PdMs: hypothesis, implementation, user validation, and improvement
• Improving and refactoring existing APIs and data models
• Improving the performance of features that handle large volumes of data
• Evaluating edge-case behavior and verifying quality and reproducibility
• ※Scope of changes to duties: yes (従事すべき業務の変更の範囲：有り)

Tech Stack

• Languages: Python (FastAPI / SQLAlchemy), TypeScript (Next.js), SQL
• Databases / Search: PostgreSQL (Aurora), Elasticsearch / OpenSearch
• Infrastructure: AWS (ECS and others), GCP, Fastly
• AI coding and review: Claude Code, Codex, Cursor, CodeRabbit
• Workflow: n8n
• Internal tools: GitHub, Slack, Notion, Asana, Google Workspace

How We Work

• We run Scrum-based agile development in small teams formed per mission, with daily scrums, sprint planning, sprint reviews, and postmortems. You also discuss the product with board and business members almost every week, so customer problems reach your design decisions directly.
• This is how we work with AI:
• AI tool spending averages more than JPY 100,000 per person per month (company-paid). Everyone can use Claude and Gemini, and we also use Codex. You can compare models in implementation and review, and build the judgment to choose which tools fit each backend task and how to verify their output
• We ask multiple LLMs to review our work, and people check the evidence before deciding what to adopt. CodeRabbit does the first pass on pull requests, so engineers can focus on design validity, data assumptions, and impact
• Engineers use AI to prepare candidate outputs. They then verify the evidence and make the final decision. We do not build AI features as black boxes, so you design features whose reasoning engineers can inspect and verify
• Once the goal and budget are agreed, the person in charge decides how to get there. Our leadership team also writes code

Ship LLM features that businesses use in their daily work

• You design and ship LLM features that structure, summarize, and interpret company data to support business decisions. You build them on SalesNow's database of more than 16 million corporate and organizational profiles, backed by 8 billion records. You own their accuracy and cost in production, which is how you build the judgment to say when an LLM feature is ready to ship and when it is not.

Shape architecture for an API with 780 PRs merged in 30 days

• As measured on 2026-09-12, 780 pull requests had been merged into the product API over the preceding 30 days. You design checks for backward compatibility, data assumptions, and LLM outputs across this volume of changes. You gain experience designing a release process that supports AI-assisted development while keeping engineers responsible for production quality. Product roles here work fully remotely, and several of our employees live in Kyushu, so you can do this from anywhere in Japan.

Design how engineers build and verify with AI

• One of our engineers runs six Claude Code sessions in parallel to carry work from investigation through implementation, verification, and release. Instructions for AI (.claude/skills/) go through pull request review like code. AI tool spending averages more than JPY 100,000 per person per month (company-paid), so you can test Claude, Gemini, and Codex against real backend work and decide which one fits each task. You decide which implementation tasks to accelerate with AI and where engineers must verify the results. You design the development process to handle a growing volume of changes while keeping quality decisions in engineers' hands.
• In two to three years, using AI will be a given. What will set people apart is the ability to redesign workflows and organizations around AI. At SalesNow, we do that every day.