Négar Tajziyehchi

Tableau Projects

Superstore Sales Agent Analytics

Accounts Receivable Dashboard

HR Analytics

Sales Analytics

Other Tableau Projects

Python Projects

Quizify (LLM, Vertex AI, Langcahin)

My Skills

Publication

About

I graduated from the University of Calgary with a Master's degree in Computer Engineering, specializing in Software Engineering. My academic journey at the university was instrumental in shaping my expertise, while my professional role as a Machine Learning Developer at AltaML provided invaluable

real-world experience. Currently, as an Associate Data Scientist at Cybera Inc., I am actively refining my data science and machine learning expertise to address and overcome practical, real-world challenges. I have a keen interest in Machine Learning and its practical implementations.

For my Streamlit dashboard book recommender project, I developed a sophisticated model leveraging TensorFlow and TensorFlow Recommenders (TFRS) to provide personalized book recommendations. The project is structured to address retrieval and ranking tasks, ensuring a two-step suggestion refinement process. Initially, the retrieval phase focuses on narrowing the vast space of possible books to a shortlist that aligns with the user's interests. This is achieved through user and book embeddings, capturing the nuanced relationships between users, authors, and book titles based on historical interactions and preferences.

A key feature of this project is its implementation within a Streamlit dashboard, offering an interactive and user-friendly interface. Users can input their details, such as user ID, preferred author names, and age, to receive tailored book recommendations. The backend model dynamically adapts to user inputs, showcasing the flexibility and responsiveness of the system.

Link to the project: Book Tensorflow Recommender System

In this Streamlit-based project, I integrated OpenAI's GPT-3.5 API to develop an AI biking guide app. The application uses prompt engineering to guide the AI in providing cyclists with personalized bike route recommendations. By specifying the AI's role and context, the app ensures relevant and focused advice on biking trails, safety, and urban cycling routes. The setup involves fetching an API key from environment variables, emphasizing security and best practices in software development.

The technical stack includes Streamlit for the web interface, allowing for an interactive chat experience, and OpenAI's API for generating dynamic responses. Custom CSS enhances the UI, ensuring a visually appealing and user-friendly application. This project illustrates the effective use of AI for personalized recommendations and the application of prompt engineering to direct AI responses within a specific domain.

Link to the project: Bike Route Guide- Chatbot

Project Overview:

As part of a dynamic team of four, I contributed to developing a Safety Recommender System for a leading oil and gas company. This project aimed to enhance workplace safety by providing actionable safety recommendations, leveraging advanced data extraction and analysis techniques.

Challenges Addressed:

Data Extraction from Complex Documents: We tackled the challenge of extracting text from approximately 10 PDF documents, each containing around 150 pages of critical safety reports.
Integration of Diverse Technologies: Our approach integrated multiple technologies to ensure comprehensive data extraction and analysis.
Development of a User-Friendly Recommender Dashboard: We aimed to make safety recommendations accessible and actionable for the company's personnel.

Solution Pipeline:

Data Extraction Pipeline:
- Initial Extraction: Utilized Azure Form Recognizer and PymuPDF for robust text extraction from PDF files. This dual approach ensured high accuracy and completeness in text data retrieval.
- Data Combination and Structuring: The extracted text data from both sources were compared, combined, and structured into a JSON file, creating a standardized format for further processing.
- Image Extraction: Employed PymuPDF to extract images from the reports, organizing them in a folder structure based on the Year and page numbers, facilitating easy access and correlation with textual data.
Dataset Creation: Developed a comprehensive dataset encapsulating incident reports alongside corresponding image paths. This dataset served as the foundation for our analysis and recommendation generation.
Recommender Dashboard Development: Designed and implemented a user-friendly dashboard to deliver safety recommendations. This tool allows company personnel to easily access and implement safety measures based on historical data and incident analysis.

Technologies Used:

Microsoft Azure
Azure Form Recognizer
PymuPDF
JSON for data structuring
Dashboard development technologies (Streamlit)

Skills Demonstrated:

Team collaboration and project management
Advanced data extraction and processing
Analytical thinking in developing practical safety solutions
Technical proficiency in utilizing Microsoft Azure, Azure Form Recognizer, PymuPDF, and data structuring techniques

Here, I present three dashboards that leverage Tableau's data visualization capabilities to uncover insights into customer profiles, e-commerce trends, and departmental performance. These dashboards illustrate my skill in using Tableau to convert complex data into clear, actionable insights, demonstrating my analytical prowess and ability to inform business strategy through data-driven visual storytelling. Each project is tailored to answer specific questions, showcasing a range of analytical techniques and visualization methods.

What does the customer look like in different regions? (link)

Geo-Visualization of Brazilian E-commerce Sellers: Payment Insights (link)

How different departments performing? (link)