Portfolio Projects
KnowBot
- Description: Knowbot is designed to be an intelligent assistant that can seamlessly interact with your organization’s internal knowledge base. Whether it's through voice or text queries, this assistant taps into your existing documents, policies, research papers, and other content to provide accurate, up-to-date, and contextually relevant answers. It blends retrieval-augmented generation (RAG) for data sourcing and a large language model (LLM) for sophisticated reasoning and answer generation.
- Tech: Python, Openai, Gradio, Whisper, LLM
Multi-Agent Financial Advisor System
- Description: This project is a production-ready multi-agent AI financial advisory system designed to deliver holistic, personalized financial guidance. It integrates investment advisory, tax optimization, and retirement planning into a unified intelligent platform.
The system leverages:
Specialized AI agents coordinated by an Orchestrator Agent
Real-time market intelligence
Advanced Retrieval-Augmented Generation (RAG) with hybrid search
Structured financial data pipelines
Educational visual generation using OpenAI Images
The advisor interprets a user’s full financial profile (income, assets, liabilities, goals, risk tolerance) and produces coherent, actionable financial plans while resolving conflicts between investment growth, tax efficiency, and long-term retirement objectives.
- Tech: Python, Gradio, Langchain, Vector-Database, Semantic Search, RAG
AzureQbot
- Description: A modern React frontend for a Knowledge Base Chatbot hosted on Azure. Features a sleek chat interface with Markdown rendering, avatars, message timestamps, dark mode, and seamless integration with a Python backend API.
- Tech: React, FastAPI, Azure, QA, Knowledgebase
AI Brochure Generator
Description: In this project, we built a highly efficient brouchure generator using LLM. We also added the option of translating the generated brochure. The solution process can be divided into three main stages. To build our graphical interface we used gradio wich is an open-source Python framework that simplifies the creation of interactive web interfaces for machine learning models, APIs, or any Python function
The first step is to build a website scrapper that can retrieve the content of a given url website. After scrapping the website, we'll send the useful website content to an LLM model. The LLM model will generate a brochure by summarizing and extracting the useful information. We choose Chatgpt and Claude to do this task. The final step is to send the generated brochure to another LLM for translation into the desired language.In our context, we decided to translate it into French.
Tech: LLM, Anthropic, Gradio, BeautifulSoup
Deal Finder
- Description: In this project we built an advanced Multi agent that subscribes to RSS feeds, check for a new opportunity deal (product), when the Multi agent finds a good deal it returns a notification containing the title, description, price, and url of the products it found.
- Tech: LLM, RAG, Langchain
Insurance Chatbot
- Description: In this project, we built an AI insurance ChatBot to help respond to customers. In order to build our AI insurance ChatBot, we used an LMI that was combined with a knowledge base using the technique known as RAG. The first step is to use a knowledge base containing information about the insurance company. We extract important context from this knowledge base. We will then vectorize the data from the knowledge base to produce better queries for our data. To prevent the LLM from making a mistake and returning the wrong answer to the client, we instruct the LLM to return the answer if it doesn't exist in the knowledge base.
- Tech: LLM, GPT, RAG, ChromaDB