Level: Senior

Job Field: IT, Data, DevOps

Employment Type: Full Time

Contract Type: Permanent employment

Salary: from 67.400 € Gross/Year

Location: Vienna

Working Model: Hybrid, Onsite

Job Summary

In this role, you develop ML/AI systems and orchestrate LLMs with agent frameworks while implementing RAG, establishing LLMOps, and conducting evaluations to optimize performance and costs.

Job Technologies

Your role in the team

End-to-end ownership of ML/AI/agentic use cases: framing to production.
Implement MCP and orchestrate LLMs/tools with agent frameworks (LangChain, LlamaIndex, Semantic Kernel, OpenAI Assistants), including function calling, fallbacks, and guardrails.
Build and optimize RAG; evaluate for precision/recall and latency.
Establish LLMOps/MLOps (experiments, versioning, CI/CD, registries, monitoring, incident response) and ensure reliability, safety, and compliance (prompt-injection defenses, content filtering, policy, red teaming, quality gates).
Conduct rigorous offline/online evaluations (backtests, time-series CV, A/B tests, shadow/canary deployments), monitor drift/impact, and optimize performance/cost (latency, throughput, rate limits, batching, streaming, caching) using usage/cost dashboards.
Interact with Business Partner to Define requirements.

This text has been machine translated. Show original

Our expectations of you

Qualifications

Expert Python skills; SQL/PySpark; Spark/Databricks and expertise in DS/ML/LLM.
Hands-on experience with agent frameworks and tooling (LangChain, LlamaIndex, CrewAI or similar), prompt engineering, function/tool calling, and evaluation harnesses.
RAG expertise: embeddings, vector stores, retrieval, and evaluation approaches.
Ability to translate business requirements into production-grade technical solutions and to manage stakeholder relationships; fluent English required, German is an advantage.

Experience

5+ years delivering ML/AI systems in production; experience in financial services is an asset.
Practical experience with GenAI models and pipelines and agentic AI (architectures, planning, memory, tool use, multi-agent orchestration).
LLMOps / MLOps experience: MLflow or similar, feature stores, data/prompt versioning, CI/CD (GitHub Actions, Jenkins), Docker, orchestration tools (Airflow, Prefect).
Experience designing evaluation and validation frameworks: backtesting, out-of-sample testing, A/B, drift detection.