AI Engineer

  • Python
  • LangGraph
  • LangChain
  • PostgreSQL
  • Fast APi
  • MCP
About the product

We are building a cloud platform that helps small US discrete manufacturers see where they stand against their peers and what to improve next. At its core is an agentic AI product, where software agents research, score and advise. We have made a start, and the agent capabilities will grow through several phases of development. You will build the intelligence at the center of it.

Role summary

You will be a hands-on AI engineer with a strong backend background, who designs and builds the agentic workflows, the AI worker that runs them, and the evaluation that keeps their output trustworthy. You will work closely with a senior backend engineer, a frontend engineer and a cloud specialist, and you will decide how AI features are built, tested and kept within budget.

Key responsibilities
  • Design and build agentic workflows in Python, with multiple steps, tool use, state and retries.
  • Build and use MCP (Model Context Protocol) servers, so agents can call backend tools and APIs in a controlled way.
  • Build and own the AI worker: a background program that picks up long running jobs from a queue, tracks progress, recovers from failures and respects cost limits.
  • Design prompts and structured outputs that code can validate, and improve them through measured testing.
  • Build the evaluation for AI output: expert reviewed test sets, an LLM judge, and tracking how closely the judge agrees with human experts.
  • Put model calls behind a clean interface, so providers can be swapped and most testing runs without calling a live model.
  • Track cost, latency and quality for every model call, and set limits and alerts.
  • Integrate AI features with the backend modules and APIs, together with the backend engineer.
  • Work with the frontend engineer on how AI output, progress and failures appear on screen.
  • Write tests, review code, and document how each AI flow works and how it is evaluated.
  • Keep up with fast moving AI tooling and recommend what is worth adopting.
  • Take part in technical decisions and mentor other engineers as the team grows.
Primary skills (must have)
  • Four to seven years of software development with strong Python 3, and at least two years building applications on large language models.
  • Agentic workflows with LangGraph, LangChain or a similar framework: multi-step flows, tool calling, state handling and error recovery.
  • Tool integration with MCP (Model Context Protocol): connecting agents to backend tools and APIs through MCP servers and clients.
  • LLM APIs and prompt design: calling model APIs, writing and versioning prompts, and structured outputs validated with Pydantic.
  • Evaluating AI output: building test sets, using an LLM as a judge, comparing it with human judgement, and regression testing prompts.
  • Backend engineering: FastAPI or similar, REST APIs, PostgreSQL with SQL, asynchronous Python, and background jobs with queues and workers.
  • AI coding assistants: hands-on use of at least one, such as Claude Code (preferred), GitHub Copilot or Cursor, for building, reviewing and testing code.
  • Engineering process: automated testing with pytest or similar, SDLC, Agile, a Git based team workflow, and CI/CD.
  • At least one AI feature or product shipped to production, not only prototypes.
Secondary skills (we expect most of these)AI toolkit

You know the main model providers (such as Anthropic, OpenAI and Amazon Bedrock) and can compare them for quality, cost and speed. You understand retrieval basics, embeddings and vector search (for example with pgvector), and the Model Context Protocol (MCP), including how to build your own MCP servers. You can feed spreadsheets, PDFs and web pages into AI flows.

Observability, cost and reliability

You trace and log AI flows with a tool such as Langfuse or LangSmith. You track cost per call, apply budget caps, and handle rate limits, timeouts, retries and caching. You test with recorded or fake model responses so tests are cheap and repeatable.

Data and cloud

You are comfortable with PostgreSQL and data modelling, and you have deployed containerised Python applications to AWS. You know Docker, S3, secrets handling and IAM, and you can read basic CI/CD pipelines.

Data engineering and analytics

You understand data pipelines, ETL and ELT patterns, and the basics of analytics engineering, and you can work with BI tools and engineers. You see how operational data and AI output later feed dashboards and historical analysis, even though your own focus is on agent driven solutions.

Responsible AI and security

You understand guardrails, prompt injection, privacy of client data, source attribution, and where a human should review AI output before it reaches a user.

Good to have
  • Evaluation tools such as Ragas, DeepEval or Promptfoo.
  • Web research and crawling tools, such as Playwright.
  • Reading and writing Excel files with openpyxl.
  • Infrastructure as code (Terraform or AWS CDK).
  • dbt or a BI tool such as Power BI or Metabase.
  • Experience in B2B SaaS, or in manufacturing or industrial domains.
Experience and education
  • Four to seven years of software development experience, with at least two years on LLM based applications.
  • BE, BTech or MCA in computer science or a related field from a recognised institution, or equivalent proven experience.
  • AI or data certifications are preferred, not required.
Location

Remote or hybrid. The company has no office yet. Our customers are in the US, so some overlap with US hours may be needed.

Facebook Twitter
Sounds like a match?

Want to turn your ideas into brilliant applications?

Talk to our Experts
Quarks

Want to give wings to your career?

Apply Now

Stay up to date with insights from Quarks!

    How can I help you?