Project Overview
Developed an LLM-powered system to simulate how tech recruiters evaluate resumes and to help job seekers assess how well their profiles align with specific job descriptions. The tool automates job-fit scoring, fact-checking, and structured feedback using prompt engineering and document querying via LlamaIndex for intelligent resume screening.
Project Overview
Designed a context-aware RAG pipeline using a custom late chunking retriever to improve document-level question answering. Combines embedding-based retrieval with LLMs to reduce hallucination and enhance factual accuracy in QA tasks.
Project Overview
Developed a predictive system to identify signs of depression and anxiety in students from textual input, using both deep learning and statistical machine learning models. The project combines BiLSTM and BiGRU architectures with traditional models (e.g., SVM, Random Forest) to analyze emotional cues in language. A web interface built with Streamlit allows real-time inference, making the system accessible for mental health screening and awareness.
Project Overview
Created a JavaScript-based script to automate CGPA calculation for CUET students by aggregating semester results directly from the student portal. Designed to address the absence of built-in CGPA functionality, enabling students across disciplines to compute their cumulative CGPA instantly and accurately.