Overview

Hi there,

Accomplished AI and Computer Science specialist, graduating with distinction from Liverpool John Moores University and holding a strong B. Tech in Computer Science from Lovely Professional University. Specialized in Machine Learning and Deep Learning, I bring a blend of academic excellence and practical experience. My expertise spans from developing advanced ML models and DL applications to executing intricate EDA/ETL and statistical analysis. With over two years of professional experience, including a programmer analyst trainee internship, I have a proven track record of working effectively with cross-functional teams to deliver high-quality solutions. My goal is to bring exceptional value to the team, overcoming any barriers with a portfolio that stands out, and showcasing my commitment and capabilities in AI and data science.

Professional Experience
My work experience so far!

---

Projects
What I've made till date

Bird species detection web application

---

• Created a Flask web application that communicates with a TensorFlow server hosted in Docker through gRPC, allowing users to upload bird images to detect bird species.
• Bootstrap and HTML/CSS is used to design the website.
• Image uploads is restricted to the following extension only (.jpg, .jpeg, .png).

See Project On GitHub

Higgs boson detection using Accelerated Machine Learning

---

• On the basis of the HIGGS dataset, the Random Forest (RF) model and XGBoost (for comparison) were used to train datasets for classifying Higgs boson signal from background signal using RAPIDS accelerated framework.
• Furthermore, a comparison has been done to determine how well each model performs on the CPU and GPU (used RAPIDS on Nvidia RTX 3090).
• In the RF and XGBoost models, the GPU is over 177 times and 300 times quicker than the CPU, respectively, allowing for significant time savings that can be leveraged to improve the model.
• In terms of accuracy, the CPU performed slightly better than the GPU in both models when the same parameter was utilized.

See Project On GitHub

Environmental sound classification for crime detection

---

• The MLP (Multi Layer Perceptron) and Random Forest model has been used to classify the environmental sound that may be used for crime scene investigations.
• The model has been saved in .h5 (1.3 MB) and.joblib (150.6 MB) formats for MLP and RF respectively.
• MLP model performs best with an accuracy of 92.73 %, whereas RF model performs best with an accuracy of 61.11 %, indicating that MLP performs better than RF on audio or time series data.

See Project On GitHub

Object detection using Deep Learning (DL)

---

• To create a DL model, Faster R-CNN with ResNet 101 model has been used in this project to train datasets to classify three bird species. Furthermore, a comparison with the SSD MobileNet model has been made to assess how this model compares.
• ReNomTAG is used to tag the images which are in the pascal VOC format. TensorBoard was used to track experiment metrics such as loss and accuracy, as well as to visualise the model graph. To build, train, and deploy bird species detection models, the Tensorflow Object Detection API is used.
• With 10 thousand steps during training, Faster R-CNN achieved above 90% of the accuracy, whereas SSD MobileNet achieved approximately 72% accuracy with 20 thousand steps.

See Project On GitHub

Contact
Reach out to me!