Summary
I'm Sayak Mondal, a passionate CSE student from VIT Vellore with a love for AI, deep tech, and building futuristic digital experiences. I enjoy merging creativity with code to solve real-world problems, and believe in making technology accessible, meaningful, and fun. From Generative AI and neural networks to web development and automation, I'm always learning, building, and pushing what's possible!
Projects
FinallyBot (2026)
Full-stack AI-powered financial assistant and quantitative trading terminal featuring JWT + OTP-secured authentication, conversational AI agent orchestration, real-time market data pipelines, and a 20 strategy quant engine with streaming execution.
Tech stack: FastAPI, Python 3.13, MongoDB Atlas (PyMongo), OpenAI (gpt-4o-mini), Google Gemini 2.5 Flash, yfinance, BeautifulSoup, Flask-SocketIO, Server-Sent Events (SSE), WebSockets, JWT (python-jose), bcrypt, pyotp, slowapi (rate limiting), React.js
Highlights: AI agent pipeline (intent → tools → LLM), real-time market analytics, financial calculators, 20-strategy backtesting engine, SSE/WebSocket streaming, MongoDB memory & scraping, production-ready REST API (44 endpoints). Created at VIT Hackathon Yantra'26 : Bidathon'26.
MedBot (2025)
Full-stack AI-powered medical chatbot delivering reference-backed responses using a Retrieval-Augmented Generation (RAG) pipeline.
Tech stack: Flutter Web (Provider, flutter_markdown), FastAPI, Python, Uvicorn, MongoDB, REST APIs, RAG architecture
Highlights: Semantic retrieval over medical knowledge base, context-aware response generation, scalable ASGI backend, web-deployable Flutter frontend.
GlobeGenie (2025)
Full-stack AI-powered global explorer generating interactive, location-driven travel guides with 3D visualization and contextual AI content.
Tech stack: React.js, Tailwind CSS, Three.js, FastAPI, Python, Uvicorn, Map/Geo APIs (GMaps API), REST
Highlights: 3D globe/map rendering, AI-contextualized travel info, modular frontend with backend for real-time geographical exploration.
EmotionCam
Real-time web-based multimodal AI application integrating facial emotion recognition, speech-to-text, tone classification, and sentiment analysis using live webcam and microphone streams.
Tech stack: Python, Flask, Flask-SocketIO, DeepFace, Vosk (offline STT), Hugging Face Transformers (DistilRoBERTa, Twitter-RoBERTa), React.js (Vite), WebRTC APIs, Socket.IO
Highlights: Real-time facial emotion detection via DeepFace, offline speech recognition using Vosk STT model, transformer-based tone & sentiment classification, WebRTC-powered live media capture, bi-directional streaming with Flask-SocketIO, privacy-first architecture (local processing, no external data transfer), dynamic dashboard with emotion bars, confidence scores & tone tags.
ResumeIQ — Smart AI Resume Analyzer
AI-powered tool to analyze resumes and suggest job compatibility scores using ML.
Tech stack: AI-powered tool to analyze resumes using ML
Image Classification Model
Classifies images from a dataset using trained ML models with a user-friendly interface.
Tech stack: trained ML models with a user-friendly interface
Sentiment + Rental + Image Classifier
Combined NLP and CV model that performs multi-task classification in one ML pipeline.
Tech stack: Combined NLP and CV model in one ML pipeline
CIFAR-10 Image Classifier & AI Agent
Built a deep learning model for CIFAR-10 image classification and experimented with AI agents.
Tech stack: deep learning model for CIFAR-10 image classification