Harsh Praharaj

About me

Information About me

Data Science masters student at University of San Francisco (Graduate in June '23) with over 3 years industry experience assisting business owners with decision making by developing world class machine learning solutions that conquer real-world business challenges. Passionate about translating business problem into a data science solution and solving it by converting complex data into meaningful stories to uncover insights. Furthermore, communicating these complex solutions in ways that business owners understand.

Download CV

My Skills

Airflow

80%

Python

90%

AWS

70%

Databricks

70%

SQL

85%

Spark

75%

My Timeline

2002 - 2014

Secondary School - Lokhanwala Foundation School

2020 - 2022

Machine Learning Engineer - Quantiphi Analytics

2014 - 2016

Senior Secondary Education - Pace Junior Science College

2022 - 2023

Data Science Master's Degree - University of San Francisco

2016 - 2020

Computer Science Degree - NMIMS University

2022 - present

Data Science Intern - SubWire

My Portfolio

Here is some of my work that I've done pertaining to Data Science/Engineering(Hover to see more).

Predicting Chess games

Designed an automated ETL pipeline using Airflow to pull data (approximately 3 million rows) of chess games from Lichess/Chess.com APIs and process it on GCP buckets. Transferred the stored data from the buckets to MongoDB clusters and connected to a databricks cluster to develop classification models.

Airflow, Spark, GCP, MongoDB

Search Engine Implementation

Implemented three different searching mechanisms to efficiently search through a large corpus of 4500 news articles.

Python, Hashtables

3D Semantic Segmentation - Colon Cancer

Developed a model to automate the detection of Tumour in a MRI. Implemented the encoder-decoder based deep learning architecture :- U-NET with Adam optimizer. Optimized the model by converting it to a TFLite Model to reduce the latency.

Keras, TFLite, NiBabel

Software Assisted Driving Aid

Developed a vision-based driver assistance system to enhance the driver’s safety with modules like Traffic Signal Classifier, Drowsiness Detection and Lane Detection.

CNNm YOLOv3, OpenCV

Demand Forecasting using Backtesting

Created an automated framework for a demand forecasting problem to choose the best-performing model on a weekly basis.

FBProphet, ARIMA, Holt's Winter, Simple Exponential Smoothing

Plant Seedling Classification

Used pre-trained VGG16, ResNet50, and InceptionV3 networks to extract bottleneck features and built a model on top of each of them to evaluate and compare the model performances.