1 day ago
Sibiu, Romania +2 moreMid Level
Responsibilities
- Design and build complex multi-agent systems and agentic architectures.
- Implement orchestration logic including state machines, graphs, retries, fallbacks, human escalation, memory, and tool use.
- Build RAG pipelines, tool-calling workflows, system prompts, and LLM-powered applications.
- Evaluate and improve agent performance through dataset curation, automatic and human evaluation, bias and safety checks, and regression testing.
- Develop agents using Google ADK and/or LangGraph and integrate libraries, vector databases, message queues, monitoring tools, and supporting infrastructure.
- Define success metrics and build evaluation suites for agents.
- Deploy and operate AI services in production with CI/CD, observability, logging, and tracing.
- Debug end-to-end failures across models, prompts, APIs, tools, and data, and document root causes.
- Collaborate with product managers and stakeholders to translate requirements into agent capabilities.
- Document system designs, decisions, and operational runbooks.
Requirements
- Bachelor’s degree in Computer Science, Engineering, or a related field.
- At least 3 years of experience as a Software Engineer, ML Engineer, or AI Engineer, including 1–2 years working directly with LLMs in real applications.
- Strong proficiency in Python, including packaging, testing, asynchronous programming, and type hints.
- Strong software engineering practices involving Git, unit and integration testing, code reviews, and CI/CD.
- Experience building and consuming REST and gRPC APIs and integrating external tools and services.
- Understanding of core ML concepts, common evaluation metrics, and deep learning fundamentals.
- Experience with at least one ML or deep learning framework such as PyTorch, TensorFlow, JAX, or scikit-learn.
- Deep practical knowledge of large language models, tokenization, context windows, sampling parameters, prompts, and prompt engineering patterns.
- Experience with fine-tuning, adapters, instruction-tuning, or RAG as an alternative.
- End-to-end experience building LLM-powered applications from prototype to production.
- Familiarity with AI safety and reliability topics including hallucinations, guardrails, content filtering, and privacy.
- Conceptual understanding of agentic frameworks and patterns such as stateful graphs, multi-agent coordination, human-in-the-loop workflows, memory, and evaluation.
- Hands-on experience with Google ADK or LangGraph and ability to read, reason about, and extend codebases using them.
- Direct production experience with both Google ADK and LangGraph is preferred.
- Experience with vector databases such as Pinecone, Weaviate, pgvector, or Chroma for RAG.
- Comfort with SQL and basic data modeling.
- Experience deploying on at least one major cloud platform and using managed services.
- Nice-to-have experience with Vertex AI, Gemini, LangChain, LlamaIndex, semantic search, RAGAS, custom evaluation harnesses, OpenTelemetry, Prometheus, Grafana, New Relic, or Datadog.
Tech Stack
Categories
About NTT DATA
NTT DATA is a Tokyo‑headquartered, publicly traded IT services firm within the NTT Group. It provides consulting, application development, system integration, cloud and enterprise application services, and managed/outsourcing services for large enterprises and public-sector organizations. The company operates in more than 50 countries and is listed on the Tokyo Stock Exchange.