You build the intelligence that runs in ArcGEN
You are responsible for our LLM pipelines, agent architectures and evaluation frameworks — from prototype notebooks to production-ready pipelines with measurable quality
Design and operation of RAG pipelines (embeddings, vector stores, re-ranking)
Architecture of agentic systems — tool-calling, MCP, sub-agent orchestration
Build and maintain LLM evaluation frameworks: metrics, benchmarks, A/B testing, quality gates
Prompt engineering and systematic prompt optimization
Develop ML models for marketing use cases (attribution, targeting, forecasting, anomaly detection)
Convert notebook explorations into production-grade pipelines
Work closely with our full-stack engineers on platform integration

Hands-on experience with modern LLM production workflows: RAG, embeddings, vector databases (Pinecone, Weaviate, pgvector)
Solid Python skills including pandas, NumPy, scikit-learn and PyTorch or TensorFlow
Experience across the end-to-end ML lifecycle: data preprocessing, training, evaluation, deployment, monitoring
Experience with systematic model evaluation: metrics, benchmarks, evaluation frameworks
German C2 and English C1
Builder mentality: you deliver production systems, not just notebooks
Developing custom agents, sub-agents or specialized coding workflows (plus)
Experience with MCP (Model Context Protocol) and tool-calling architectures (plus)
Experience with agent frameworks (Mastra, LangGraph, Vercel AI SDK) (plus)
Experience with fine-tuning, LoRA adapters or distillation (plus)
MLOps tools (MLflow, Weights & Biases, DVC) (plus)
Background in marketing analytics, attribution models or performance marketing data (plus)
Kaggle, open-source contributions or publications in ML/AI (plus)

Competitive salary with annual adjustments
Fully funded access to Claude Max, Codex, GitHub Copilot, OpenAI and Anthropic APIs, and all relevant LLM tools
Compute budget for experiments, model training and ML workloads
MacBook Pro of your choice
Remote-first, flexible working hours
Quarterly team meetups in Berlin or Dubai (DIFC AI Campus, optional)
Direct influence on product architecture at an early stage
Steep learning curve in agentic AI, LLM orchestration and AI-native engineering

AI Engineer – LLM Apps, RAG & Agentic Systems, German-speaking

Key skills