Senior AI Engineer

Deine Rolle im Team

Unsere Erwartungen an dich

Qualifikationen

• Generative AI Expertise: Good understanding of modern Generative AI techniques and foundation models, including transformer-based Large Language Models (LLMs), diffusion models, and multimodal models, as well as earlier architectures such as GANs and VAEs.
• Hands-on exposure to both API-based (e.g., Claude, GPT, Gemini) and open-source (e.g., Llama, Mistral) LLM-based solution design.
• Familiarity with agentic design patterns (e.g., ReAct, planning, reflection, tool use, human-in-the-loop) and agent frameworks such as LangGraph, CrewAI, MAF, the OpenAI Agents SDK, and Google's Agent Development Kit (ADK).
• Model Context Protocol (MCP) & Interoperability: Practical understanding of the Model Context Protocol (MCP) for standardized, secure connectivity between LLMs/agents and external tools, data sources, and systems.
• Ability to build and consume MCP servers and clients, and to work with MCP primitives such as tools, resources, and prompts.
• Awareness of related interoperability standards (e.g., agent-to-agent communication) for composing enterprise-grade agentic systems.
• Ability to design custom tools, connectors, and skills that let agents perform specialized, domain-specific tasks reliably and safely.
• Familiarity with advanced patterns such as GraphRAG and agentic RAG to reduce hallucination and improve factual grounding.
• Technical Proficiency: An overall understanding of below technologies is required : Machine learning algorithms: Linear regression, logistic regression, decision trees, random forests, support vector machines, neural networks.
• Data science tools: NumPy, SciPy, Pandas, Matplotlib, TensorFlow, Keras.
• Cloud computing platforms: AWS, Azure, GCP.
• Natural language processing (NLP): Transformer models, attention mechanisms, word embeddings.
• Computer vision: Convolutional neural networks, recurrent neural networks, object detection.
• Robotics: Reinforcement learning, motion planning, control systems.
• Data ethics: Bias in machine learning, fairness in algorithms.
• Foundation models & LLMs: GPT, Claude, Gemini, Llama, Mistral; multimodal and reasoning models; context windows, tokenization, and fine-tuning (LoRA/PEFT), RLHF/RLAIF concepts.
• LLM application & agent frameworks: LangChain, LangGraph, LlamaIndex, Semantic Kernel, Haystack, CrewAI, AutoGen.
• Interoperability & integration: Model Context Protocol (MCP), function/tool calling, structured outputs, API integration, event-driven and orchestration patterns.
• Cloud AI platforms & model hosting: Amazon Bedrock, Azure OpenAI / AI Foundry, Google Vertex AI, Hugging Face.
• Vector databases & retrieval: Pinecone, Weaviate, Chroma, pgvector, FAISS; embeddings, semantic and hybrid search, reranking.
• MLOps / LLMOps & deployment: Docker, Kubernetes, FastAPI, CI/CD; observability, tracing, and evaluation tooling (e.g., LangSmith, LangFuse); guardrails and prompt/version management.
• Responsible AI & safety: bias and fairness, hallucination mitigation, evaluation, privacy, security, and governance of AI and agentic systems.
• Solution Design: Ability to design end-to-end Generative and Agentic AI solutions, from requirement elicitation and model selection to deployment strategy.
• Familiarity with guardrails, red-teaming, and responsible deployment of AI systems in production.
• Communication Skills: Excellent verbal and written communication skills to engage with clients, articulate technical concepts to non-technical stakeholders, and work collaboratively with cross-functional teams.

Erfahrung

• Proven experience in applying these techniques to real-world problems for tasks such as text, code, image, and multimodal generation.
• Agentic AI & Orchestration: Hands-on experience designing autonomous and multi-agent systems that reason, plan, and act using tools.
• Experience building agentic workflows with memory, state management, and reliable multi-step task execution.
• Agent Skills & Extensibility: Experience extending agent capabilities through modular, reusable skills-packaged instructions, scripts, and resources (e.g., SKILL.md-style capability modules) that agents load on demand via progressive disclosure.
• Retrieval-Augmented Generation (RAG) & Knowledge Systems: Proven experience designing RAG and knowledge-grounded systems, including chunking strategies, embeddings, vector databases (e.g., Pinecone, Weaviate, Chroma, pgvector, FAISS), hybrid search, reranking, and evaluation of retrieval quality.
• Experience crafting architectures that encompass data preprocessing, RAG pipelines, agent orchestration, MCP-based tool and system integration, model integration, guardrails, and performance, cost, and latency optimization.
• LLMOps, Evaluation & Optimization: Experience operationalizing LLM and agentic applications-building evaluation harnesses and offline/online metrics for quality, groundedness, and safety; implementing observability, tracing, and monitoring; and continuously optimizing accuracy, cost, and latency.

Unser Angebot