Назад
Company hidden
4 дня назад

Software Engineering – AI Agent

Формат работы
onsite
Тип работы
fulltime
Грейд
senior
Английский
b2
Страна
China
Вакансия из списка Hirify.GlobalВакансия из Hirify Global, списка международных tech-компаний
Для мэтча и отклика нужен Plus

Мэтч & Сопровод

Для мэтча с этой вакансией нужен Plus

Описание вакансии

Текст:
/
TL;DR
Software Engineering – AI Agent (Python/TypeScript, RAG): Building AI agent applications that connect LLMs with enterprise systems, laboratory software, and scientific instruments with an accent on backend architecture, secure orchestration, and production delivery. Focus on designing agent runtimes, optimizing retrieval pipelines, integrating model APIs and MCP services, and ensuring reliability through testing, observability, and access controls.

Location: Shanghai, China

Company

hirify.global develops software and technology for scientific, laboratory, and enterprise applications.

What you will do

  • Design and deliver AI agent applications, backend services, and APIs using Python and TypeScript/Node.js.
  • Build agent runtimes with prompt and context management, tool calling, persistent state, recovery, and human approval.
  • Develop and optimize RAG pipelines covering ingestion, embeddings, vector or hybrid search, reranking, retrieval evaluation, and source-grounded responses.
  • Integrate approved cloud or local model APIs with enterprise and laboratory systems, including reusable MCP clients and servers.
  • Own production quality through testing, CI/CD, access controls, prompt-injection defenses, and monitoring of accuracy, latency, cost, and failures.
  • Lead architecture and delivery decisions, mentor engineers, and collaborate with product, science, UX, quality, and support teams.

Requirements

  • Bachelor’s degree or above in computer science, software engineering, AI, automation, data science, or a related technical field.
  • 5+ years of software engineering experience, including production development in both Python and TypeScript/Node.js.
  • Strong foundations in asynchronous programming, API design, testing, and debugging.
  • Hands-on delivery of LLM applications or agents using cloud or local model APIs and a relevant framework or SDK such as LangGraph, LangChain, Semantic Kernel, or LlamaIndex.
  • Practical end-to-end RAG experience and proficiency with at least one vector database or search platform, including retrieval evaluation, metadata filtering, and relevance troubleshooting.
  • Experience with prompt and context engineering, structured outputs, tool orchestration, state management, failure recovery, and choosing deterministic workflows when appropriate.
  • Strong API, database, and authentication integration skills; familiarity with MCP, Docker, Linux, CI/CD, observability, least-privilege access, and secure handling of model outputs.
  • Ability to own architecture and production delivery, collaborate across functions, communicate effectively in English, and support AI interfaces when needed.

Nice to have

  • Advanced agent-platform experience with long-running or multi-agent workflows, reusable MCP services, evaluation pipelines, and operational recovery.
  • Local inference or hybrid cloud/local deployments, advanced retrieval optimization, and scalable services using FastAPI, Redis, queues, Kubernetes, or OpenTelemetry.
  • Experience with scientific software, laboratory automation, instrument integration, LIMS/CDS, regulated workflows, or .NET/C# applications.
  • Experience with Angular, Electron, or React for agent workspaces and approval interfaces, plus enterprise identity and Responsible AI practices.

Culture & Benefits

  • Full-time weekly schedule with a day shift.
  • Occasional travel is required.
  • No fixed end date for the position.
  • Pay ranges vary by role, level, and location; specific pay and benefit information is provided during the hiring process.
  • Equal opportunity employment under applicable laws.

Будьте осторожны

Если работодатель просит войти в их систему, используя iCloud/Google, прислать код/пароль, запустить код/ПО, не делайте этого. Это мошенники. Обязательно жмите «Пожаловаться» или пишите в поддержку. Подробнее в гайде →