AI/ML Developer
AI/ML Developer
I design production-ready AI solutions with RAG, LLM applications, and agentic automation.
Hi! I'm Nemish Sapara, an AIML Developer focused on RAG, LLM applications, and agentic AI. I design practical AI products that combine strong model performance with clear product thinking.
A collection of tools and frameworks I use to build powerful applications
Scikit-Learn
OpenCV
React
Flutter
HTML5
CSS3
JavaScript
MongoDB
MySQL
DockerScikit-LearnOpenCVReactFlutterHTML5JavaScriptMongoDBMySQLDockerPublished work contributing to AI & Machine Learning research
This paper presents an enhanced CNN-based face swapping framework integrating multiple deep learning architectures for real-time video processing. The system utilizes a multi-stage pipeline incorporating InsightFace models, INSwapper networks for identity transfer, and GPEN-based enhancement models, achieving up to 30 FPS with high-fidelity face synthesis and temporal consistency.
Proposes an intelligent WhatsApp-based verification system leveraging AI to authenticate ration card beneficiaries in India's Public Distribution System, targeting rural accessibility, reducing fraud, and enabling real-time eligibility checks through a conversational interface.
Some things I've built that I'm proud of
AI-powered Flask web app for processing lectures — upload audio or enter text to get instant speech-to-text transcription via OpenAI Whisper, intelligent summarization via BART, and multilingual translation via Google Translate API. Features drag-and-drop upload, voice input, and dark/light theme.
A collection of end-to-end ML projects built with Jupyter Notebooks — including Diabetes Prediction, Food Delivery Analysis, and a Fraud Detection System. Covers full pipelines: data preprocessing, model training, evaluation, and visualization with real-world datasets.
A centralized cross-platform mobile app built with Flutter + Firebase for the National Service Scheme at Charusat University. Connects students, volunteers, and coordinators with role-based dashboards, event management, attendance & hours tracking, push notifications, and a communication hub.
A full-stack MERN web app bringing the timeless wisdom of the Bhagavad Gita to life. Features mood-based verse exploration, audio recitations, personal favorites & notes, daily visit streak, powerful keyword search, related verses, JWT authentication, and dark/light mode.
Real-time AI web app that detects faces from live camera input and predicts emotions instantly. Built for interactive emotion recognition with a clean, fast interface and responsive performance.
End-to-end support ticket router that classifies multilingual customer messages into Technical Support, Billing Inquiry, or Order Tracking using a FastAPI backend and an interactive React operations console.
Telegram-based multi-agent memory assistant that stores text, photo, and document memories, then answers natural-language queries using semantic vector search and RAG. Built with n8n automation, Groq LLMs (text + vision), Cohere embeddings, and Supabase pgvector.
Advanced fraud detection dashboard built for the ML InnovateX Hackathon featuring real-time transaction monitoring, batch processing with CSV uploads, and comprehensive analytics. Uses an ensemble model combining Logistic Regression and Artificial Neural Networks with 97.36% AUC score and PCA feature selection.
Enterprise-style employee helpdesk and knowledge assistant with role-aware retrieval, citations, feedback capture, audit logging, and admin workflows for document approval, tickets, and policy notifications.
Have a project in mind? Let's build something amazing together.
