Projects

This repository contains a short video explaining how modern AI systems retrieve information from large document collections.

1. Embeddings: transforming text into numerical vectors that capture semantic meaning.

2. Vector Databases: storing document embeddings for efficient similarity search.

3. Retrieval: selecting the most relevant document chunks for a user query.

4. LLM Context Generation: providing retrieved information to a language model to produce accurate answers.

These techniques are widely used in AI assistants, knowledge search systems, and enterprise chatbots.

Link: JeffreyRed/LLMOps 

This repository contains examples and experiments demonstrating key NLP techniques:

1. TF-IDF (Term Frequency – Inverse Document Frequency): classic method for measuring word importance in documents.

2. RAG (Retrieval-Augmented Generation) using OpenAI LLMs: combining embeddings and GPT to answer queries with grounded information.

Link:  JeffreyRed/LLM_RAG: Understanding LLM and RAG 

Build an intelligent customer support agent that can autonomously handle billing issues by: Understanding customer complaints in natural language, Retrieving relevant billing information from databases, Making decisions about refunds and corrections, Taking actions through API calls, Providing clear, empathetic responses to customers.

Key Difference from Traditional Chatbots: This system doesn't just generate text responses—it actively performs actions like checking databases, issuing refunds, and solving problems end-to-end without human intervention.

Link:  JeffreyRed/Billing-Agentic-AI-customer-support 

Here are two scenarions one for task manager, and email organizer. For email I decided to implement TF-IDF for vectoring the emails. Finally , RAG is implemented to add knowledge. In additin a comparison script comparison.ipynb

Link:  JeffreyRed/Agentic_ai_Task_Completion_Text_Agent: Project Idea: “Task-Completion Text Agent” 

An agentic AI system that creates personalised language lessons based on Bible study using OpenAI's ChatGPT.

Link:JeffreyRed/Language_Study_and_Bible_Study 

Evaluation of interference in an heterogeneous network and interference management in MATLAB simulation. 

It is a WPF un C# that can communicate with BG96 Quectel module via serial sending AT commands, and getting the coordinates from an external GPS or in built computers. and finally it generates a log file with the data.

• NB-IoT Drive Test Tool.

Research project regarding the performance of the new technology for IoT connectivity NB-IoT over LTE. During this project, I coded the boards to get a reliable connectivity between the BG96 module to the eNB. After that, the data acquired have been analyzed for research purpose. In this moment, we are working in the PSM (Power save mode BG96 module) and Low power mode (SMT32 Microcontroller) to measure the power consumption. 

• NB-IoT LTE project at Tallinn University of Technology.

The final outcome of the winter school was the development of an environment where the Digital Twins approach could show the important of have a replica of the real process. For this project, I could develop a program to follow mu hands move and compare then with the real world move done in Arduino car. 

• Twin Industry Using Arduino and Unity. Winter School.

An android App that shows the location of a devices that is connected to a Dora Network. This app was developed for IoT devices.