Projects

These are the projects I have worked on and my clients are happier.

Passionate AI enthusiast dedicated to crafting innovative solutions that blend creativity and expertise, showcased through a portfolio of impactful projects that speak volumes.r eyes linger here, and see if you can get a feel for our signature touch.

Real Time AI Face Landmark Detection App with Tensorflow.JS

I built a Real Time AI Face Landmark Detection App with Tensorflow.JS

Facial landmark recognition allows you to detect a number of different points on your face that together make up your eyes, mouth, ears, nose, and so on.
Check it out here! Project

Text Classification -Transformed toxic comment classification using deep learning and TensorFlow

Aiming for a healthier online environment!

Accomplishments:

Engineered a high-performance model using Bidirectional LSTM layers.
Achieved impressive precision, recall, and accuracy metrics in toxicity prediction.
Deployed an interactive Gradio interface for real-time toxicity scoring.

Check it out here! Project

FashionGAN: AI-Generated Fashion with Generative adversarial networks
(GANs)

What I Did:

Imported essential dependencies and data for the project.
Visualized and preprocessed the Fashion MNIST dataset, making it suitable for GAN training.
Designed and built a GAN model from scratch.
Trained the GAN model for 20 epochs (Note: Longer training is recommended for optimal results).
Monitored and saved generated fashion images during training using a custom callback.
Reviewed and visualized the model's performance.
Developed a custom callback for monitoring and saving generated images.
Gained experience in GAN training and model evaluation.

Check it out here! Project

Text Detection and Recognition with EasyOCR-OpenCV

🌟 What I Did:

I developed a robust system for text detection and recognition using the powerful EasyOCR library and OpenCV. This project allows the automatic extraction of text from images.

📌 Benefits and Accomplishments:
✅ Achieved high-precision text localization by applying filters and edge detection techniques.
✅ Implemented contour analysis to isolate and extract the text area from the image.
✅ Leveraged EasyOCR to accurately recognize and extract text content from the cropped region

Check out Project Here!

Fine-Tuned the BERT Model for Sentiment Analysis

I leveraged state-of-the-art NLP techniques to fine-tune a BERT model for sentiment analysis. Here's what I accomplished:

🧐 Project Overview:

1) Utilized the Hugging Face Transformers library and Hugging Face Datasets to streamline my NLP workflow.
2) Fine-tuned the bert-base-uncased model for binary sentiment analysis on the IMDb dataset.

🌟 Benefits and Accomplishments:

1) Achieved exceptional results in sentiment analysis, providing accurate sentiment classification for text data.
2) Improved model accuracy by optimizing hyperparameters, including a learning rate of 2e-5.
3) Developed a high-performance model capable of processing and classifying text data efficiently.

Checkout Project Here!

Built a Chatbot Script Generator with LangChain 🦜️🔗 - Streamlit! 🚀✨

The Script Generator with LangChain is a groundbreaking tool that leverages advanced language models and AI to automate video scripting, providing developers with an efficient and creative content generation solution. It's a fusion of technology and innovation exciting for coders and AI enthusiasts.

Key Achievements:

User-Friendly App Framework: The code sets up an easy-to-use interface using Streamlit.
Automated Title and Script Generation: It uses LangChain and OpenAI's LLMs to generate YouTube titles and scripts automatically.
Dynamic Prompt Templates: Users can specify topics and utilize Wikipedia research for customization.
Memory Functionality: The code stores and retrieves chat history to maintain context.
Interactive User Interface: It offers an engaging and user-friendly visual interface.

Checkout Project Here!

Water Quality Predictions using Machine Learning

🧪 Applied machine learning models like Logistic Regression, KNN, SVM, Decision Trees, Random Forest, and XGBoost to predict water potability.

Accuracy I have Obtained in these Models:

Logistic Regression: Achieved an accuracy of 75%.
K-Nearest Neighbors (KNN): Achieved an accuracy of 82%.
Support Vector Machine (SVM): Achieved an accuracy of 79%.
Decision Trees: Achieved an accuracy of 88%.
Random Forest: Achieved an accuracy of 93%.
XGBoost: Achieved an accuracy of 87%.

Ensuring clean and safe drinking water is critical. This project paves the way for more accurate water quality predictions, which can have a significant impact on public health.

Check out Project Here!

Spotify Recommendation Engine

I employed data science techniques and ML algorithms for personalized music recommendations. Achievements:

Explored/Prepared data: Analyzed song/artist info, and extracted features.
Feature Engineering: Normalized variables, one-hot encoding, TF-IDF for genres.
Spotify API: Connected to retrieve user playlists, music info.

Skills: Data exploration, feature engineering, API integration.

Iris Flower Classification

The Iris Flowers dataset, which I worked with, consists of numeric attributes and is divided into three species:

Iris Setosa
Iris Versicolour
Iris Virginica

To accomplish this task, I utilized various libraries and packages such as pandas, numpy, matplotlib, seaborn, and more.

With their help, I imported and analyzed the dataset, gaining a comprehensive understanding of its attributes and structure.

Stock Market Prediction & Forecasting Using Stacked LSTM

Utilized data science techniques and deep learning for stock market prediction.