Graph-Augmented RAG for Telecom

Knowledge Graph and Graph-Augmented RAG for Telecom Intelligence.
Ericsson is building a new R&D team in Massy, France, covering many cutting-edge technologies such as AI/ML, cloud, and 5G advanced/6G technologies. Within this new lab, Standards & Technology unit, is part of Development Unit Network's global Standards & Technology organization. At Development Unit Networks, Standards & Technology secure technology leadership in Radio Access Networks (RAN) by actively driving New Concepts, Standardization, SW- and HW Research, Architecture and Testbeds. Technical focus includes 5G evolution, IoT, Digital twins, Automation / Machine Learning and Security.
As a part of this new young and talented team of researchers, we are looking for motivated interns that will help us grow and engage in our activities on generative AI for telecom.
Context:
Telecommunication systems are defined through thousands of pages of technical specifications, standards, and white papers describing protocols, interfaces, procedures, and architectural principles.
Accessing and reasoning over this knowledge is difficult because the information is:
• Deeply hierarchical (documents → sections → clauses → parameters)
• Highly inter-referenced (cross-references across specifications)
• Evolving across releases and organizations
Traditional Retrieval-Augmented Generation (RAG) approaches flatten the information space into independent text chunks, losing the structural hierarchy and relationships inherent in telecom documentation.
Hierarchical RAG (Hi-RAG) is a new paradigm that aims to preserve and exploit document structure by combining:
• Hierarchical retrieval (parent/child context navigation)
• Graph reasoning (relations among entities)
• LLM generation grounded on multi-level evidence.
This internship explores Hi-RAG as a foundation for telecom knowledge intelligence, bridging document structure and semantic graphs to improve factual accuracy, explainability, and reasoning depth.
Research Questions:
• How can we represent telecom knowledge in a multi-level hierarchical-graph structure suitable for retrieval?
• What retrieval strategy (top-down, bottom-up, or hybrid) yields the best factual grounding?
• To what extent does hierarchical context reduce hallucination and improve explainability compared to standard RAG?
• How can the graph hierarchy be used to guide adaptive context selection under token budget constraints?
Objectives:
In this internship we are looking for talented students to help us to design, implement, and evaluate a Hi-RAG framework capable of understanding and reasoning over telecom domain documentation.
• Extract and structure telecom documents into a hierarchical format.
• Build a knowledge graph capturing telecom entities and their relationships.
• Link graph entities to document hierarchy for integrated semantic understanding.
• Develop a hierarchical retrieval mechanism that navigates between parent, child, and sibling sections.
• Combine hierarchical retrieval with LLM reasoning to create a Hierarchical RAG (Hi-RAG) system.
• Design prompts and context assembly methods optimized for structured, citation-based answers.
To be successful in the role you need to have:
• Basic understanding of telecommunication systems and network architectures (4G/5G or similar).
• Interest in knowledge representation, document intelligence, and retrieval-augmented generation (RAG).
• Experience with Python programming and data processing libraries.
• Experience with cloud-based AI platforms, preferably AWS and Amazon Bedrock (or similar cloud LLM services).
• Interest in working with graph technologies (Neo4j, NetworkX, or RDF-based tools).
• Curiosity about Generative AI, Large Language Models, and their applications in telecom.
• Qualities of fast learning, critical thinking, autonomy, and teamwork.
• Willingness to work in an inclusive, research-oriented, and multicultural environment.
• Fluent English language skills in both writing and conversation.
• French language skills are a plus.
Join our Team
About this opportunity:
Ericsson is searching for a skilled and diligent Local Administrator to perform a variety of administrative activities within our local legal entity. As a Local Administrator, your role will be crucial in supporting various processes across different functional areas, especially where local representation is insufficient. You will be tasked with ensuring optimal cross-functional efficiency within our Market Areas and adapting to specific requirements informed by stakeholder demands, local laws, and regulations. Adherence to our Code of Business Ethics, including responding to Anti-Bribery and Corruption red flags, is fundamental to this role. If you are adept at handling diverse responsibilities and working collaboratively, we invite you to apply.
What you will do:- Administer local financial functions such as payment processing and withholding tax activities.- Oversee and manage travel and expense activities.- Maintain and handle records retention tasks.- Oversee company card applications and ensure timely payments.- Manage local benefit administration, including social and medical insurance coverage.- Assist with Vehicle Fleet Management activities like vehicle ordering, data management, and condition inspection.- Perform Purchase-to-Pay activities including invoice verification and processing.- Handle IT-related tasks such as device distribution and repair coordination.- Actively engage in compliance and improvement activities.- Assist with Customer Invoices and Cash Collection activities.- Lend support to Sourcing and Sales activities as needed.
The skills you bring:- People Services.- Process Improvement.- Cash Management.- Microsoft Office 365.- Withholding Tax.- English language.- Credit Card Applications.- Purchase Order Preparation.- Document Management.- Invoicing.- Compliance Requirements.- Automation Anywhere (RPA Software).- Administrative Law.- Payment Processing.- Archiving.