Experience Real
Projects

AI real life Projects and Blogs
and the potential of Artificial Intellegence.

Scroll Down

TECHNICAL SKILLS

Languages and Libraries :

C/C++,python, MATLAB/Simulink, Numpy, Pandas, Tensorflow, Keras, Matplotlib, OpenCV, Scikit-learn, Seabon.

Data Structures and Algorithms :

C++.

Web-Development :

CSS, HTML, JavaScript, Bootstrap.

Developer Tools :

Visual studio code, Jupyter NoteBook, Google Collab, MATLAB, Pycharm

PROJECT EXPERIENCES/TECHNOLOGY STACK

I have extensive experience working with a diverse set of technologies, including: Machine Learning, Deep Reinforcement Learning, Deep Learning, CNNs, ANNs, Saimese Neural Netwok, Deep NLP, RNNs , Seq2Seq Model, KivyMD for application development, AutoEncoders and Restricted Boltzman Machine for Rating and Recommenders system.

ACHIEVEMENT

Smart India Hackathon 2023 - Winner(Intra College), Team Leader & AI Developer – Led a victorious team, showcasing AI development skills and strategic leadership in a prestigious nationwide hackathon.

What are
Projects?

Computer Vision

Convolutional Neural Network, Digital image processing

–Developed a real-time sign language decoding model using MediaPipe Holistic Key Points for palm and pose detection, Tensorflow/Keras, and Stacked LSTM layer for enhanced effectiveness.
–Created a facial recognition application with a Siamese network model using TensorFlow/Keras for face authentication and opencv Haar–Casscade for face dectection integrated it into a Kivy application, achieved precision and recall (1.0).
–Created an application for responsive media content interaction through hand–gestures and computer vision, utilizing OpenCV, numpy and PyAutoGUI to develop a gesture-controlled video player via digital image processing in python.

ChatBot Kivy Application

Deep Natural Language Processing

–Built a Chatbot using Deep Natural Language Processing (DNLP) , enhancing conversational interactions.
–Implemented and Trained a Seq2Seq DeepNLP model using the Cornell Movie Dialogs Corpus dataset.
–Employed stacked LSTMs and Attention Mechanisms for effective context understanding and response generation.
–Developed a Kivy Application seamlessly integrated with the model, enabling real–time, human–like interaction.

MuJoCoAI

Deep Reinforcement Learning, Q–Learning

–Implemented a Deep Q–Network and A3C model to enhance robust decision–making in Non–Deterministic environments.
–Created a real–time customizable environment and car, for self-driving using Kivy and Lunar Lander from OpenAI Gym.
–Implemented Twin Delayed DDPG model to find the optimal policy for Markov Decision problem and achieved high
cumulative rewards in complex MuJoCo environments such as Ant, Half Cheetah, and Humanoid

Recent Blog

Computer Vision

Facial Recognition with Siamese neural network

Facial Recognition is something you we can see everywhere in our day to life, like in phones and other equipment for security purposes. In humans also faces this phenomenon a lot or I would say nearly their whole life from childhood to their last breath....[Read More]

Computer Vision

Sign Language Detection using Action Recognition

Built a Real-time Action Detection model to decode sign language, empowered by LSTM layer for enhancing effectiveness. •Extracted MediaPipe Holistic Key Points for palm and pose detection using the Mediapipe library for training and testing. Developed a robust deep neural network using Tensorflow & Keras, utilizing a Stacked LSTM layer to effectively process and analyze the pattern from the sequences of detected holistic key points for real-time sign language decoding...[Coming Soon..]

Deep Reinforcement Learning

MuJoCo environment solver AI

Implemented a Deep Q–Network and A3C model to enhance robust decision–making in Non–Deterministic environments. Created a real–time customizable environment and car, for self driving using kivy and Lunar Lander from OpenAI Gym. Implemented Twin Delayed DDPG model to find the optimal policy for Markov Decision problem and achieved high cumulative rewards in complex MuJoCo environments such as Ant, Half Cheetah and Humanoid.... [Coming Soon..]

Featured In

Goal is to design efficient self-learning and intelligent software to automate predictive models and contribute to advancements in artificial intelligence..

Twitter

LinkedIn

GitHub