Hi, I'm Anmol Sharma

Voice-guided form assistant that makes government paperwork accessible to elderly and non-tech-savvy users. Upload any PDF, answer plain-language spoken questions, and receive a fully-filled, ready-to-submit document no typing required. Built a vision-LLM pipeline (Qwen2.5-VL) on AMD hardware to extract and interpret every fillable field, with OpenAI Whisper for transcription and ElevenLabs for natural TTS delivery.

Founded and led the development of Q-Stream, an open-source media player. Architected the full-stack application, utilizing Python, PyQt5 for the GUI, Streamlink for video ingestion, and Flask for backend modularity. Grew the project into a collaborative open-source initiative featured in Devscript Winter of Code (DWoC).

Built a CNN-based system to detect cyberattacks by converting network traffic into GAF images. Achieved 99.8% accuracy on the CIC-IDS-2017 dataset using models like ResNet18, MobileNetV2 and custom CNN.

Developed an NLP-powered application using Streamlit that automates resume screening. The engine leverages semantic search with Sentence-Transformers and cosine similarity to rank resumes, and includes a generative AI component using a Hugging Face T5 model to summarize each candidate's fit.
Live app • Code

Designed and developed a chatbot using Python and Google Dialogflow to provide users with comprehensive college information and answer frequently asked questions. Built intent-based natural language processing models to accurately interpret user queries and provide relevant responses.