Senior Backend Engineer, ML Platform

必須

• 5+ years of professional software engineering experience designing, building, and operating web APIs and microservices in Python or comparable languages
• Track record with high-traffic, low-latency applications through efficient API design and application-level performance tuning
• Contract and failure design experience for external system integrations including API schema, error taxonomy, timeouts, retries, and fallbacks
• Strong test design and quality assurance background with experience embedding automated testing into the development process
• Cloud application development experience on AWS using containerized environments and CI/CD pipelines
• Ability to collaborate and build agreement across roles to drive technical specifications with SRE, machine learning engineers, and external stakeholders
• Business level Japanese (equivalent to JLPT N2 or above)
• Basic business level English (equivalent to TOEIC 700 or above)
• If you do not have a qualification equivalent to TOEIC 700 or above, you will be required to take a company-designated test during the selection process.

歓迎

• Knowledge of machine learning integration and MLOps, including calling models on platforms like Amazon SageMaker and designing for inference latency
• Understanding of data platforms and pipelines like S3, Glue, and Databricks to inform technical specs and incident analysis
• Fluency reading and writing IaC with Terraform to propose infrastructure changes via pull requests
• Experience in financial services, credit, or payments domains with exposure to auditability, traceability, and advanced security like mTLS, OAuth, or JWT
• Leadership experience as a tech lead or scrum master with authority over architecture, code quality, and process standardization

求める人物像

• Wants final accountability for product behavior with first-line responsibility for the correctness, speed, and availability of a credit API designed through to the last failure case
• Drawn to converting the uncertainty of machine learning into the certainty of an API by designing an interface that behaves deterministically despite inference variance, latency, and model updates
• Capable of standing at the boundary between roles to design the collaboration itself through clear ownership while picking up problems that fall between roles
• Thrives on shaping systems during phases of genuine uncertainty and enjoys making the decisions that move projects forward when specifications and architecture are undecided
• Eager to build infrastructure-scale products for the AI era by leading the launch phase of Digital Bank combined with shared credit evaluation using AI development tools such as Claude Code

使用ツール

Overview

• This is a dedicated _backend engineering_ role owning the _application layer of a credit microservice (a private REST API)_ used across our Digital Bank and group-company fintech products. The microservice combines aggregated data produced by Databricks, machine learning model endpoints on SageMaker, and credit calculation logic inside the service itself to deliver credit information to the product side. In short, you will _turn probabilistic, uncertain model output into deterministic API behavior a financial product can safely depend on_ .
• _Success is clearly defined: ensuring the FY2027 Digital Bank launch succeeds, from the standpoint of the credit microservice._ Your mission is the architecture, implementation, and quality assurance that work backward from it.

Mission Priorities

• 1. Design, build, and assure the quality of a credit microservice (private REST API) capable of supporting the FY2027 Digital Bank launch
• 2. Stabilize machine learning model integration, and secure resilience through latency budgeting and fallback design
• 3. Establish development standards, architecture, and test strategy, and drive the team's technical decision-making

Responsibilities and Duties

• As the primary application owner for the microservice, you will take the initiative within our scrum team and:
• _Design and implement the microservice architecture:_ a high-performance, low-latency service in Python and FastAPI, built for transaction volumes at the scale of tens of millions of users, including service boundaries and data ownership.
• _Design API contracts and reach agreement with stakeholders:_ OpenAPI schemas and error taxonomies, versioning and backward compatibility, and the process for changing contracts with Digital Bank and group-company stakeholders.
• _Bring models and credit logic into production, with resilience by design:_ the client-side interface for SageMaker endpoints, latency budget allocation, and timeout, retry, circuit breaker, and fallback behavior.
• _Implement credit domain logic with correctness and explainability:_ credit limit calculation with exhaustive boundary and failure cases, idempotency, reproducibility (linking input data, model version, and logic), and audit trail design.
• _Design for performance around data supply:_ how aggregated data in object storage is loaded, freshness management and consistent cutover, caching strategy, and asynchronous/parallel model calls.
• _Own quality, test strategy, and instrumentation:_ unit, contract, and load testing, regression processes for model swaps, structured logging and traceability, and SLI-driven instrumentation.
• _Lead technically and set standards:_ architecture selection, code review, and improvement and standardization of our development process.

Technology Stack

• Backend / API: Python (FastAPI), Amazon API Gateway
• Infrastructure / Cloud: AWS (ECS Fargate, S3, CloudWatch)
• ML Serving: Amazon SageMaker (Model Endpoint)
• DevOps / CI/CD: GitHub Actions, Docker, Terraform
• Data Platform (integration target): Databricks, AWS Glue, S3
• AI Development Tools: Claude Code and similar
• Communication / Project: Slack, Notion