Sales conversations hold a wealth of untapped insight — and AI is the key to unlocking it. The Agent team at amptalk is building intelligent, autonomous systems that listen, understand, and act on behalf of sales and customer success teams. From real-time coaching agents to automated follow-up workflows, we are pioneering AI-native products that fundamentally change how businesses sell.
As a Software Engineer on the Agent team, you will design and build AI-powered agent systems that sit at the intersection of large language models, event-driven architectures, and product engineering. You won’t just integrate AI — you’ll shape how our agents reason, decide, and deliver value to thousands of users every day.
This is a high-autonomy, high-impact role for engineers who are excited about pushing the boundaries of what LLM-based systems can do in production.
Technologies
- AI/LLM: OpenAI SDK, prompt engineering & optimization
- Backend: Node.js, TypeScript, Effect (functional TypeScript)
- Cloud: AWS (Lambda, DynamoDB, SQS, SNS, SES, CDK)
- Integrations: Google APIs (Gmail, Calendar), Microsoft Graph API, Zoom API
- Testing: Vitest
- CI/CD: GitHub Actions, pnpm
Note: The above technologies reflect our current stack. The Agent team has the autonomy to make technology choices that best serve our AI-native product goals. You will be involved from the architecture and technology selection stage for new agent products.
Responsibilities
- Design, build, and operate AI agent systems powered by large language models (LLMs), including prompt engineering, output parsing, and evaluation pipelines
- Architect event-driven workflows (Lambda, webhooks, queues) that enable agents to act autonomously and reliably at scale
- Collaborate closely with product, sales, and CS teams to translate business needs into intelligent agent behaviors
- Own the full lifecycle of agent products — from prototyping and prompt tuning to production deployment and monitoring
- Evaluate and integrate new AI capabilities (models, APIs, tooling) to continuously improve agent performance
- Contribute to engineering best practices through code reviews, documentation, and knowledge sharing
Requirements
- 4+ years of software engineering experience in product development
- Hands-on experience building applications that integrate LLMs (e.g., Claude, GPT) — including prompt design, tuning, and output evaluation
- End-to-end engineering experience from design and development through to production operation
- Solid backend engineering skills (Node.js/TypeScript preferred)
- Experience with event-driven or serverless architectures (Lambda, webhooks, message queues)
- Strong communication skills and comfort working with cross-functional stakeholders
- Proficiency in English
- Proficiency in Japanese (native level not required; working proficiency is a strong plus given our Japan-focused market) Guideline: JLPT N2 level or above
- Passion for continuous improvement and staying current with the rapidly evolving AI landscape
Nice to haves
While not specifically required, tell us if you have any of the following.
- Experience designing, building, and operating architectures on AWS
- Familiarity with RAG (Retrieval-Augmented Generation) patterns, vector databases, or embedding-based search
- Experience building multi-step agent workflows (e.g., tool use, chain-of-thought, planning)
- Knowledge of SOLID principles, Clean Architecture, or Domain-Driven Design (DDD)
- Test-Driven Development (TDD) experience
Compensation
¥7,000,000 ~ ¥10,000,000 annually.
