Shaurya Chawla

I’m a Computer Science enthusiast with a strong interest in building impactful, real-world solutions that go beyond just code. Alongside my technical skills in software development and AI systems, I bring a strong focus on communication, influence, and collaboration—skills that help me work effectively across teams and understand user needs deeply.

Outside of tech, I’m deeply connected to music. I play drums and have hands-on experience in audio engineering, which has trained my ear for detail, timing, and creative expression. This blend of analytical thinking and artistic discipline shapes how I approach problem-solving—structured, but creative.

I’m particularly drawn toward non-coding and hybrid CS roles where technology meets people—such as product, AI strategy, technical consulting, or developer relations—where communication and technical understanding come together to create impact.

MY RESUME

PROJECTS

VIEW ALL

Automatic Speech Recognition (ASR):

Developed an Automatic Speech Recognition (ASR) system using Deep Learning techniques. Implemented a Recurrent Neural Network (RNN) with attention mechanism to transcribe speech into text. Utilized Mel-Frequency Cepstral Coefficients (MFCCs) for feature extraction and trained the model on LibriSpeech dataset. Achieved high transcription accuracy and optimized performance with Beam Search decoding.

AI CHATBOT
(full stack)

I developed a session-based chatbot application using FastAPI, PostgreSQL, and a lightweight frontend to deliver a seamless conversational experience. The system begins with user onboarding, collecting essential details such as name, email, and phone number with proper input validation to ensure data integrity. Each user interaction is managed through a unique session ID, enabling personalized and continuous conversations. The chatbot stores complete chat history in a PostgreSQL database, allowing for efficient data retrieval and tracking. The backend is built using FastAPI and SQLAlchemy to provide a robust and scalable REST API, while the frontend features an interactive interface with dynamic chat bubbles for an engaging user experience. Overall, the application supports both structured questioning and free-flow conversation, demonstrating a practical integration of backend engineering, database management, and user-centric design.

Document Reader Chatbot (RAG-based)

I built a Retrieval-Augmented Generation (RAG) chatbot that enables users to query PDF documents and receive accurate, context-aware answers strictly grounded in the document content. The system processes uploaded PDFs by splitting them into meaningful text chunks, which are then converted into semantic embeddings using SentenceTransformers. When a user submits a query, it is similarly embedded and matched against the stored vectors using cosine similarity to retrieve the most relevant context. This context is then passed to an OpenAI-powered language model to generate precise, fact-based responses while minimizing hallucinations. The application is deployed through a Flask web interface, offering a simple and interactive user experience. Overall, the project demonstrates practical implementation of modern RAG pipelines, combining natural language processing, vector search, and large language models to build reliable, document-driven AI systems.

AI- voice Bot
(Twilio based)

I developed a real-time AI voicebot that enables natural, human-like conversations over phone calls using Twilio Media Streams and OpenAI-powered speech and language models. The system captures live audio from user calls through WebSockets, processes it by converting μ-law audio into WAV format, and performs speech-to-text transcription to understand user intent. The transcribed input is then passed to a large language model to generate intelligent, context-aware responses, which are converted back into speech using text-to-speech synthesis and streamed back to the caller in real time. Built with FastAPI and optimized for low-latency performance, the voicebot supports continuous audio streaming, handles noisy environments, and adapts to multilingual input while maintaining English responses. This project demonstrates end-to-end integration of telephony, real-time audio processing, and conversational AI to create a seamless voice-based user experience.