
Hi, I'm Jahanzeb!
Welcome to my portfolio.
About Me
I have over 3 years experience in data science & machine learning and a 1st class honours degree in Mechanical Engineering. My academic journey reached its pinnacle with an innovative final year project which involved using AI to optimise a manufacturing process, reflecting my passion for leveraging advanced AI technologies. Building on this foundation, I honed my skills and gained practical experience during my time at Vodafone which has equipped me with a unique skill set, allowing me to seamlessly bridge the gap between engineering principles and the transformative power of data.
As you explore my portfolio, my work will show my commitment to innovation, problem-solving and delivering impactful solutions.
VIKI: The Vodafone Internal Chatbot
Skills Demonstrated
While working at Vodafone, I found myself using private Large Language Model (LLM) chatbots more and more to assist me which gave me the idea to develop a chatbot specifically for Vodafone employees to scan internal documents and receive answers for queries relating to these documents. I presented this idea to some peers and worked with them to develop a prototype on my personal comuputer which I then presented to the innovation approval board at Vodafone to get granted resources to develop this into a cloud app for the entire company, shown below. The chatbot was named the Vodafone Internal Knowledge Interpreter, VIKI for short, and I received a 'Future of Vodafone' award for developing it.
Predicting Video Game Sales
Skills Demonstrated
This project's aim was to develop an app for video game developers and publishers which used a deep learning model to predict the sales of their products. Initially, to explore and analyse a dataset I obtained online, I developed a dashboard which revealed that the data was incomplete and so, to rectify this, I wrote a Python script to scrape the missing data from Wikipedia and Metacritic to ensure the training dataset for the final model was robust and complete. Furthermore, the dashboard was useful in highlighting which features in the dataset had an effect on the video game sales and needed to be included for training and, after experimenting and testing different hidden layer and neuron configurations, I trained a neural network model from the PyTorch library to have an MAE of 0.008. Finally, to allow others to benefit from my project, I used Streamlit in Python to develop a webapp of the dashboard which also used the trained model to give users predictions for the sales of their video games.
Optimising the MW-PECVD Process in Industry
Skills Demonstrated
Microwave-Enhanced Chemical Vapour Deposition (MW-PECVD) applies thin film coatings onto substrates to give desirable hardness and elastic properties. However, tuning the parameters of coating machines to ensure specific properties are obtained while also ensuring temperatures remain below max. operating temperatures can be challenging. In the final year of my engineering degree, I carried out this project to develop an optimisation tool which used a multi-layer perceptron (MLP) deep learning model to recommend optimum parameters for industries to obtain their desired coatings while taking into consideration economic and sustainability issues. After submitting my project, I was encouraged by my supervisor to make the tool available and so I developed a web app using Streamlit in Python which I have received positive feedback on from student and faculty at University of Leeds.