RA-KT: Cognition and Metacognition Tracing

What if I can make students/learners solve their assignments while training them for metacognition and after the assignment was submitted/graded, give them a per-solution performance report including student's awareness, outlook and skill mastery percentage as performance metrics

Ever felt like interviews are a game of hide and seek, and you’re always ‘it’? What if you had a personal AI-powered bot to help you prepare? Dive into my journey of creating an interview bot using OpenAI’s GPT-3.5 and Streamlit.

This project implements a SQL Writer model based on the GPT (Generative Pretrained Transformer) architecture. The model is trained to generate SQL queries given a context. The implementation closely follows the official GPT paper.

A BERT model trained to predict a student's confidence levels while/before solving a problem.

Checkout Jordan Peterson speak about IITs in Hindi (kinda gibberish due to lack of GPU). For a detailed breakdown, please check the following links:

github - https://github.com/dibyanshuchatterjee/Youtube-Video-Language-Translator

I tried creating a Drowsiness Detection system that uses Eye-Aspect-Ratio (EAR) to detect if the user is drowsy or not. If the EAR is found below a defined threshold, an audio is played (Micheal from The Office screaming "No, No God No, NOOO"). 

https://github.com/dibyanshuchatterjee/Realtime-Drowsiness-Detection

Remember the good ol' days of decoding assignments through strategic Googling? 🕵️‍♂️ Then came #llms, and suddenly telling a student's work from a machine's became the ultimate brainteaser!

I tried exploring that balance through LLM's chain of thoughts. 

What if I could tell the sentiment of what has been spoken? I guess #lstm models are apt enough to satisfy the use case :D

This LSTM model takes in an audio file and does a pretty good job of analyzing the sentiment.

https://github.com/dibyanshuchatterjee/LSTM-for-call-transcript-sentiment-classification-over-call

Hey,
* What if I could cheat the way I type on WhatsApp using Computer Vision?

* What if I could send a few generic messages through a few gestures?
* What if I could switch to a virtual keyboard to write sentences?

https://github.com/dibyanshuchatterjee/Virtual-Keyboard-With-Hand-Gesture-Classification

Ever wonder what Raju from Hera Pheri feels when Babu Bhaiya tells him "Has na halkat"?

Well, if you did, here is a frame-by-frame emotion detection of Raju's emotions using Deep Learning.

https://github.com/dibyanshuchatterjee/Emotion_detection_deep_learning_parent