LLM Backend Engineer

We’re building a new product from scratch - focused on pragmatic, production-ready LLM-based agents.
The research and architecture phase is happening. Now we need engineers who will help turn those decisions into a real, scalable system.
This is not about blindly playing with prompts.
This is about building backend services, APIs, evaluation pipelines, and infrastructure that make agentic systems reliable in production.
We’re tackling a real-world problem with a clear path to market, backed by SafeQ Cloud - and you’ll also help shape and improve the in-house tools that power the product.

What you’ll actually work on

• Designing and implementing backend services (Python or .NET)


• Integrating LLM vendors (OpenAI, Anthropic, Google…)


• Working with LangChain stacks, evaluation tooling (e.g., DeepEval)


• Building and tuning RAG pipelines, vector databases


• Connecting (and building) MCP servers


• Making sure what we build is testable, observable, and deployable


• Helping evolve the architecture as we move from prototype to production

You won’t just receive tasks. You’ll participate in refinements and technical discussions - but the focus is clear: build it well and ship it.

Your toolbox

• Strong backend experience (medior+)


• Solid engineering in Python or .NET (C#)


• Experience building APIs and production services


• Comfort with cloud and DevOps basics (CI/CD, containers, infrastructure awareness)


• Interest in LLMs and how to use them responsibly in real systems


• Czech and English language

Bonus if you’ve worked with:

• RAG, vector databases


• Agent frameworks


• Microservices architecture


• Evaluation pipelines for LLM outputs


• Kubernetes / GitHub Actions / CI/CD


• PostgreSQL, Redis, S3

How we work

• We experiment - but we also ship.


• We brainstorm - but we also commit to decisions.


• We build as a team, not as isolated contributors.

This is still a greenfield product in a stable company. You’ll influence the architecture - but you’ll also help turn it into a robust system that customers actually use.