RISHIKESH JHA
Future AI & ML Engineer | Data Scientist
Future AI & ML Engineer | Data Scientist
I'm passionate about building AI-powered applications that tackle real-world challenges, with a special interest in running AI models on mobile devices.
My interests include:
Education: Completed +2 at Nepal Police School, Sanga.
Background: Coding in Robotics team of my school.
Let's connect to explore how technology can solve your problems! Find me on LinkedIn.
- Rishikesh Jha
Sklearn • PyTorch • Transformers • LangChain
MicroPython • Docker • MongoDB • Arduino
This project aims to classify emotions and severity levels from audiovisual data using the RAVDESS dataset. It features a dual-branch model architecture, with a fine-tuned 3D ResNet (r3d-18) for video frames and a fine-tuned VGG16 for audio spectrograms. The system extracts frames and spectrograms to perform multi-label classification, and is optimized for small batch processing in low-resource environments.
This project is a simple implementation of a Convolutional Neural Network built from the ground up using Python and NumPy, without relying on deep learning libraries. It demonstrates the core concepts and workings of CNNs for educational purposes.
This project implements a binary image classification model to distinguish between images of dogs and cats using a Convolutional Neural Network (CNN). Built with PyTorch, the model processes image data efficiently, achieving high accuracy on the Kaggle Dogs vs Cats dataset.
This project implements the classic LeNet-5 architecture using PyTorch for handwritten digit classification on the MNIST dataset. The model achieves an impressive 98.85% test accuracy, showcasing its effectiveness in image classification tasks.
A machine learning project to predict customer churn using various classification Decision tree and Random Forest. I explore about Encoding categorical variable, HyperParameter Tuning Using GridSearchCV.
Built a basic neural network from scratch using Python to classify handwritten digits (MNIST dataset). Implemented core components including forward/backward propagation, activation functions (ReLU, Softmax), gradient descent optimization and categorical cross-entropy loss.
The IoT-enabled Smart Poultry Farm is a project designed to automate and optimize poultry farming operations using modern technology.
We have developed 3 simulations using Unity and Arduino for Google Cardboard VR as a prototype, for training with virtual reality. They are controlled using a modified gun which is made as the controller. If further development is done, it can give more realistic experience.
Interested in collaborating? Get in touch.
Let's connect and collaborate!