Paarth Manchanda

B.Tech in Mathematics & Computing

Delhi Technological University (DTU)

2025 – 2029

DacSaab

Built an offline-capable healthcare assistant for rural health workers using quantized LLMs, XGBoost-based triage models, and multimodal AI pipelines supporting text, voice, and medical image inputs. Implemented local-language speech processing, risk-level classification, and structured patient summarization optimized for low-resource Android and desktop environments.

AI-Interview

Developed a conversational mock interview platform using a fine-tuned Mistral 7B model with QLoRA, enabling domain-specific interview simulations, grammar-aware feedback, and contextual conversational flows. Integrated Faster-Whisper and CTranslate2 for low-latency speech-to-text transcription, while designing customizable interview pipelines with Flask-based deployment support.