MLOPS ENGINEER (AWS Bedrock/LLM/Agentic AI) - UN-client | 40-day initial | Fully Remote

Posted 18 hours 34 minutes ago by infom consulting GmbH

Contract
Not Specified
Other
Roma, Italy
Job Description

SENIOR MLOPS ENGINEER (AWS Bedrock/LLM/Agentic AI) - UN OPPORTUNITY - REMOTE (50/100%, 40 days initial)

For our UN client, we are looking for an experienced:

  • Senior MLOps Engineer - AWS/Generative AI

  • 100% remote, supporting a UN agency in Rome

  • Start: To be agreed

  • Two complementary AI projects, each listing 40 days; total billable effort to be confirmed

  • 50% per project; one consultant may cover both projects at 100% allocation

  • At least 80% overlap with 08:30-17:00 CEST

TASKS

  • Productionize LLM, RAG and specialist agent workflows for data analysis, visualization, analytical report drafting and quality assurance.

  • Package and deploy workflows using Amazon Bedrock/AgentCore and approved orchestration patterns.

  • Implement prompt, agent and model configuration management with versioning, promotion, rollback and traceability across development, QA and production.

  • Operationalize knowledge-base ingestion, indexing, synchronization, retrieval and re-ranking using S3, vector stores and Bedrock knowledge services.

  • Build an advisory AI layer for data-quality findings, explanations, root-cause guidance and correction recommendations.

  • Implement structured output schemas, validation, grounding and evidence presentation, preserving deterministic validation and mandatory human review.

  • Ensure AI cannot approve or reject datasets, modify production records or autonomously trigger writeback.

  • Create evaluation datasets and regression tests for accuracy, grounding, consistency and unsafe or unsupported recommendations.

  • Apply Bedrock guardrails against prompt injection, sensitive-data exposure and unauthorized tool or data-source access.

  • Monitor usage, cost, latency, errors, retrieval quality and workflow success, with thresholds and alerts.

  • Log required AI inputs, outputs, configuration versions and operational metadata; implement fallback behaviour when services are unavailable.

  • Implement model upgrade and deprecation procedures, release approvals and rollback controls.

  • Optimize compute consumption, scalability and performance.

  • Provide evaluation reports, configuration inventories, runbooks, audit-ready documentation and knowledge transfer.

SKILLS

  • Minimum 4 years of hands-on AWS project experience.

  • Experience with Amazon SageMaker/Bedrock, AWS Step Functions, ECS and AgentCore.

  • Practical experience productionizing LLM, RAG or agentic workflows on Amazon Bedrock or comparable governed enterprise platforms.

  • Strong Python programming and modern API development.

  • Knowledge of Docker, Kubernetes, Kubeflow and MLflow.

  • Experience with CI/CD tools and DevSecOps practices.

  • Infrastructure as Code using Terraform, CloudFormation or equivalent.

  • Model monitoring, observability and performance optimization.

  • Knowledge of data engineering tools such as Spark, Airflow and Kafka.

  • Experience operating MLOps life cycles across development, QA and production environments.

  • Enterprise security, privacy, data governance and Responsible AI.

  • Ability to produce audit-ready documentation and transfer knowledge to technical and business stakeholders.

  • Relevant Bachelor's/Master's degree in Computer Science, Data Science, Engineering or a related field.

  • Strong written and spoken English.

International competition - please provide a competitive daily rate in EUR.

Direct applicants only - no agencies or intermediaries, please.

TO DO - to apply, please send:

  1. A 3-4 sentence introduction explaining how your experience matches this specific role.

  2. Your CV in Word/DOC format , clearly highlighting your relevant hands-on AWS, Bedrock and production LLM/RAG experience.

  3. Your all-inclusive daily rate in EUR for fully remote work.

  4. Your availability/notice period and earliest possible start date 

  5. Your capacity for one project at 50% or both projects at 100% allocation.

  6. Country of residence and confirmation of working-hours overlap with Rome time.

Deadline: ASAP - early applications encouraged.

Please send your application to (see below)