agentic AI engineer

NDA; Dublin; On-site; EUR 150,000 - 190,000; Full time

Job summary

NDA is creating global autonomous AI systems that plan, reason, use tools, and act for millions of users. This contract role focuses on designing and deploying production-grade agentic architectures with planning, memory, orchestration, and multi-agent collaboration.

You will optimize inference, build RAG pipelines, and develop evaluation frameworks to ensure reliability, safety, and scalability across real-world datasets.

Qualifications

• Proficient Python in production-grade systems.
• Hands-on with LLM-powered agents and orchestration layers.
• Delivery of production RAG systems with vector databases.
• Knowledge of LLM inference, batching, and GPU memory trade-offs.
• Experience with prompt engineering and function/tool calls.

Responsibilities

• Design, build, and deploy agentic AI systems with planning and multi-agent collaboration.
• Architect RAG pipelines for large real-world datasets to reduce hallucinations.
• Optimize LLM inference for latency, throughput, and cost at scale.
• Build evaluation frameworks for reliability, safety, and quality.

Skills

Python

LLM agents

Distributed systems

Prompt engineering

Cloud & MLOps

Tools

LangGraph

LlamaIndex

AutoGen

CrewAI

vLLM/TensorRT-LLM

Job description

Описание

A global technology company is building autonomous agentic AI systems that plan, reason, use tools, and act on behalf of millions of users.

Задачи

• Design, build, and deploy production-grade agentic systems with planning, tool use, memory, orchestration, and multi-agent collaboration
• Architect RAG pipelines for large, real-world datasets, including chunking, embeddings, hybrid retrieval, reranking, and grounding strategies to reduce hallucinations
• Optimize LLM inference for latency, throughput, and cost through batching, caching, quantization, speculative decoding, and serving at scale
• Build evaluation frameworks for agent reliability, safety, and quality, including offline evaluations, LLM-as-judge, red-teaming, and online experimentation
• Fine-tune, adapt, and prompt-engineer foundation models where it has measurable impact
• Partner with research, product, infrastructure, and trust and safety teams to turn ideas into shipped features
• Improve engineering quality through design reviews, mentoring, and knowledge sharing

Требования

• Strong software engineering fundamentals and production-quality Python
• Hands-on experience building LLM-powered agents using frameworks such as LangGraph, LlamaIndex, AutoGen, or CrewAI, or a custom orchestration layer
• Proven delivery of production RAG systems, including vector databases and retrieval evaluation
• Working knowledge of LLM inference and serving, including vLLM, TensorRT-LLM, Triton, KV-cache, batching, and GPU memory trade-offs
• Experience with prompt engineering, function/tool calling, and structured outputs
• Solid cloud and MLOps skills, including AWS, GCP or Azure, Docker, Kubernetes, CI/CD, and observability
• A pragmatic, product-minded approach to deciding when to build, buy, or cut scope
• Будет плюсом: TypeScript, Go, Java or C++; fine-tuning (LoRA/QLoRA, RLHF/DPO, distillation); multi-agent systems, MCP or tool-protocol design, and agent safety/guardrails; operating AI systems at very large scale; open-source contributions, publications, or a strong public technical profile; an MSc/PhD in Computer Science, ML, AI or a related field, or equivalent hands‑on experience

Условия

Contract role