Chatbot to act as an assistant to engineering teams
Διπλωματική Εργασία


Foundation large language models (LLM), such as GPT-4 and Llama2, can answer general knowledge questions and even exhibit reasoning abilities. However, they lack context specific knowledge that can be found in business organizations internal documents. The information contained in the foundation models can be extended by using context material and so called Fine Tuning or retrieval augmented generation (RAG). [2]

Following this reasoning we want to develop a chatbot that shall act as an assistant to engineering teams that install, configure, customize and provide technical support for Intracom Telecom’s BigData platform. The chatbot shall use open source LLM and Retrieval Augmented Generation (RAG) to answer user queries about the specific knowledge base that comprises of the documentation corpus of the BigData platform and the database of Service Requests (SRs) from customers. The scope of the project includes the design, implementation, testing, evaluation and optimization of the chatbot using algorithms and open source frameworks that provide automation of the tasks.


  1. Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks
  2. Chatbots in Knowledge-Intensive Contexts: Comparing Intent and LLM-Based Systems,
  3. Build a contextual chatbot application using Knowledge Bases for Amazon Bedrock
  4. Large Language Models as Optimizers
  5. RAG Evaluation

Επιθυμητές γνώσεις

Επικοινωνία: Νικόλαος Ευθυμίου (

Παναγιώτης Τσανάκας (