Developing a production-ready retrieval-augmented generation (RAG) based conversational AI assistant that provides personalised immigration advice to empower people to live and work wherever they want. Responsibilities include:

• Creating a RAG pipeline using langchain and ingesing the chunked and parsed data scraped from official government websites
• Enabling personalized advice using user database
• Establishing a evaluation pipeline

Developed a data pipeline for predicting battery drainage in electric vehicles and analysing influential features to analyse the necessary sensors for data collection. Responsibilities include:

• Cleaned and pre-processed the raw data and implemented two data architectures: segmenting and bucketting models

Analysed and accounted for the systematic error in centroiding algorithms like Centre of Mass (CoM) to increase the accuracy of finding the centre of star images. Responsibilities include:

• Worked on error-predicting algorithms like the Extreme Learning Machine with Bat algorithm as an optimiser (BA-ELM), 1D Gaussian Fitting, and Fast Gaussian Fitting; BA-ELM algorithm increased the accuracy of the CoM algorithm by 40%.
• Analysed image smoothing algorithms like Savitzky-Golay Filters and their effect on CoM accuracy.
• Evaluated star tracking algorithms to predict the centroid locations for faster extraction of stars from an image.