Portfolio
Hi! I’m Dwarkanath Prabhu. I work on machine learning and web development. Here are some of my projects sorted by topic.
Index
- Computer Vision
- Natural Language Processing
- Neural Networks
- Classical Machine Learning
- Web Development
- Certifications
Computer Vision
Skin Cancer Classification
- Description: Classified images of skin lesions into different types of skin cancer using a ResNet50 architecture pretrained on ImageNet with a precision of 87.5%
- Language: Python (Jupyter Notebook)
- Libraries: TensorFlow, numpy, matplotlib
- Code: GitHub Repo
Handwritten Digit Recognition
- Description: Identified the number (0-9) from images of handwritten digits as provided by the MNIST dataset with 99.1% accuracy
- Language: Python (Jupyter Notebook)
- Libraries: TensorFlow, numpy, matplotlib
- Code: Kaggle Kernel | GitHub Repo
Art generation with Neural Style Transfer
- Description: Created a stylized version of an image in the style of an impressionist style painting using a VGG19 architecture pretrained on ImageNet.
- Language: Python (Jupyter Notebook)
- Libraries: TensorFlow, PIL, matplotlib
- Code: Notebook
Face Recognition
- Description: Verify whether an image belongs to a known person i.e. calculate whether the L2 norm of encoding of new image and that of person’s image in the database are below a threshold. The econding is done using a pretrained OpenFace model
- Language: Python (Jupyter Notebook)
- Libraries: Keras, OpenCV
- Code: Notebook
Natural Language Processing
Tweet Language Model
- Description: Building a character-level language model based on my tweets to generate new tweets
- Language: Python (Jupyter Notebook)
- Libraries: Keras, Tweepy
- Code: GitHub Repo
Jazz music generator
- Desctiption: Generated a new jazz solo of 20 seconds using an LSTM network trained on a corpus of jazz music
- Language: Python (Jupyter Notebook)
- Libraries: Keras
- Code: Notebook
Eliminate bias from word vectors
- Desctiption: Identified bias in GloVe representation of words using cosine similarity and removed gender-based bias
- Language: Python (Jupyter Notebook)
- Libraries: Keras
- Code: Notebook
Neural Networks
Titanic Kaggle Competition
- Description: Predicted survival of a passenger on the RMS Titanic using information available about them using neural networks with an accuracy of 80% (top 10% of Kaggle).
- Language: Python (Jupyter Notebook)
- Libraries: TensorFlow, scikit-learn, matplotlib, seaborn
- Code: Kaggle Kernel | GitHub Repo
Building your Deep Neural Network - Step by Step
- Desctiption: Built a general L-layer neural network and implemented forward and back propagation from scratch using numpy
- Language: Python (Jupyter Notebook)
- Libraries: numpy, matplotlib
- Code: Notebook
Classical Machine Learning
Anomaly Detection
- Description: Applied Principal Component Analysis and Hotelling’s $T^2$ test on a dataset to identify out-of-control samples and remove them till only in-control samples were left.
- Language: Python (Jupyter Notebook)
- Libraries: numpy, matplotlib
- Code: GitHub Repo
Radar Signal Quality
- Description: Predicted the quality of a signal from a system of radars based on its attributes using SVM, Random Forest and LDA with a cross-validation accuracy of 92%.
- Language: R
- Libraries: e1071, randomForest, MASS
- Code: GitHub Repo
Web Development
Twitter Clone
- Description: A Twitter clone developed in Flask - available in English and Spanish
- Language: Python
- Libraries: Flask, Flask-SQLAlchemy, Bootstrap, Flask-WTF, Google Translate
- Code: GitHub Repo
- Live Version:
Personal Website
- Description: My personal homepage with information about me and links to my social profiles.
- Language: HTML, CSS
- Code: GitHub Repo
- Live Version: https://dwarkanath.xyz
Personal Blog
- Description: This blog! Made using Hugo site generator and written in markdown
- Language: go, HTML, CSS
- Code: GitHub Repo
- Live Version: https://dwarkanath.xyz/blog
FreeCodeCamp Projects
- Description: Single-page projects to learn HTML, CSS as part of FreeCodeCamp certification
- Language: HTML, CSS
- Code:
Week In Memes - Static Site
- Description: Static site + blog to keep track of viral phenomenon in India
- Language: go, HTML, CSS
- Code: GitHub Repo
- Live Version: https://weekinmemes.com
Certifications
Click on the image for link to the certificate on the certifying authority website.
Google Cloud Essentials
Deep Learning Specialization
Responsive Web Design
Author:
Dwarkanath Prabhu
Hi, I’m Dwarkanath Prabhu! I’m a software engineer, currently working at WorkSpan
I like reading and writing about programming, startups, soccer (Arsenal FC) and politics.