Developed a question-answering tool using Corrective-RAG (CRAG) to improve answer accuracy through self-reflection and grading of retrieved documents & Integrated a multimodal retriever to extract relevant images from PDF and PPTX files.

Utilized K-means clustering to group similar sections and reduce dimensionality, enabling efficient summarization system, capable of processing documents with 1000+ pages in less than 60s & Implemented a cumulative approach to reduce the time required for summarizing added documents & Leveraged Gemini’s multi-modal capability to build a multi-modal RAG that enables users to chat with PDFs.

Designed and developed an efficient Indexer that leverages NLP techniques to store and preprocess HTML pages,minimizing overhead and redundancy, and enabling fast data retrieval & Built a robust Query Engine that efficiently retrieve relevant results from the indexed data.

Conducted statistical analysis and hypothesis testing to gain insights into the factors affecting employee turnover & Built an early warning system that accurately predicts the probability of an employee quitting and assigns a risk level to each employee.

Applied classical computer vision techniques to build a fast and real-time license plate detection & Built an OCR model to be used in license plate recognition.

Implemented an efficient scheduling algorithm using data structures and OOP concepts to minimize response time and waiting time for the CPU.