Developed a deep learning framework to distinguish between human-written and AI-generated essays in the academic domain. The model leverages a combination of linguistic features, including readability metrics, stylometric patterns, and lexical diversity, to improve classification accuracy. By using a fine-tuned autoencoder for embeddings and a feed-forward neural network, the system effectively identifies the source of the text. This approach is designed to be robust against adversarial modifications and can be applied to detect AI-generated content in academic writing.

Exploring RAG, I built a small project called RAG Buddy. It answers queries based on the content of your uploaded PDF documents, ensuring no hallucinations. The process is simple: upload a document, the app extracts and chunks the text, embeds it into vectors using sentence-transformers/all-MiniLM-L6-v2, and retrieves the most relevant pieces to answer your question using an LLM. It also keeps the model safe from prompt injection, since it only replies using what’s actually in your document and nothing outside it.

A tool to identify potential inaccuracies or "hallucinations" in the responses generated by language models, comparing outputs against provided source material (like PDFs).

It’s a tool that simplifies database exploration by bridging the gap between natural language and SQL queries. Imagine asking your database:"Show me all customers who purchased in the last month."and instantly receiving the answer No technical expertise required! Powered by CodeLlama, Flask, and SQLite. CRM Genie transforms the way users interact with data, whether they’re sales reps, managers, or data enthusiasts.

Anemia Detection, Check whether you are anemic or not! How does it work? Just fill out the values of your blood attributes, and the model will do the rest! Latest Feature: Forget manually entering values one by one! Just upload a pic of your blood test report, and it will automatically extract the required features from the image and give you an instant prediction. How does it predict? A Random Forest Classifier trained on blood attributes does the Prediction.