Profile Image

Ameya Kshirsagar

About Ameya Kshirsagar

Ameya Kshirsagar is a B.Tech in Information Technology graduate from Symbiosis Institute of Technology with gusto to discover value-added aspects of Computer Science and Data Science, competency to take on every learning challenge. His research interest lies in NLP and machine learning applications for interdisciplinary applications ranging from finance, Healthcare, Environment, petroleum, and sports.

Email:ameyark28@gmail.com

Degree:Bachelor of Technology

Phone::D

City:Mumbai

CV

Skills

Python (Programming Language)
C++
Machine Learning
Data Science
Artificial Intelligence (AI)
Research
Deep Learning
Data Analysis
Natural Language Processing (NLP)
Mendeley

Certifications

The Science of Success: What Researchers Know that You Should Know Verify
Introduction to Google SEO Verify
Machine Learning Verify
Introduction to AWS Identity and Access Management Verify
Understanding the Impact of Deepfake Videos Verify

Achievements and Awards

Pro-Tech 2020 National Project Exhibition
Distinction - Science
1st Open State Level Martial Art Championship
Maharashtra State Level Karate Championship

Education

2016-2020

Bachelor of Technology in Information Technology

Symbiosis Institute of Technology, Symbiosis International University

● Member of Google Developer Student Club.
● Feliciated at ProTech 2020.
● Student Mentor.

Experience

May 2020 – Present

Research executive

Pandit Deendayal Energy University - Formerly (PDPU)

Working on Machine Learning, Artificial Intelligence and NLP applications in healthcare, environment, finance, and petroleum industry.

● Implemented machine learning model geothermal micro-seismic event classification and subsurface temperature prediction.

● Performed Exploratory Data Analysis for Higher education trends for Gujarat, India. ● Performed data analysis and prediction of JSC, VOC, FF, and Efficiency (%) for organic photovoltaic cells.

● Implemented machine learning models for efficiently predicting COVID-19 trend patterns.

June 2021 – August 2021

Machine Learning Engineer

Omdena

● Successfully collaborated to build healthcare solutions to predict cardiac arrest due to pulseless electrical activity and asystole employing machine learning for 'Transformative.ai' with AUC = 0.91 and accuracy = 97.05%.
● Lead a group of 11 members to accomplish a sub-task.
●Collaborated on an open-source to implement artificial intelligence techniques and satellite-image analysis to reduce freshwater wastage in Egypt.

June 2021 – August 2021

Data Science & Business Analytics intern

The Sparks Foundation

● Completed tasks with efficiency ranging from peer-reviewing, data wrangling to modeling to learn the pipeline of machine learning projects: data collection, data cleaning, data visualization, modeling, and prediction.

July 2021 – September 2021

Data Science Intern

LetsGrowMore

● Successfully implemented several machine learning and NLP models for applications ranging from stock forecasting, next-word prediction, etc., to contribute to the organization.

January 2021 – August 2021

Delegate

Harvard Project for Asian and International Relations (HPAIR)

● Selected as one of the delegates from 60+countries for the HPAIR conference, which features world-class speakers and guests to foster mentorship, networking, and guidance opportunities for delegates.

January 2019 – July 2019

Data Science apprentice

Persistent Systems

● Improved and automated insurance claiming process with image preprocessing, OCR (PyTesseract), and entity extraction to successfully cover end-toend solutions and aspects of the online claim-reimbursement system with more than 96% accuracy

Recent Projects


GitHub

EDA for Higher education trends for Gujarat, India

Collected data from a local firm, extracted meaningful insights from data to determin the hidden pattern of students going for higher education.


Flipkart customer review analysis

Extracted meaningful insights from Flipkart reviews employing topic modeling, Latent Dirichlet Allocation model.


Reinforcement Learning enabled game

Implemented Deep Q Network (DeepMind) to imitate players for game playing for snake game.


Textual question answering system

Employed deep learning model (DistilBERT) for question answering system for precise and faster rate of response.


Chat-Bot for food ordering application

Built food ordering assisting neural network-retrieval-based chat-bot with intent classification.


Text summarizer

Employed NLP, spaCy to build extractive text summarizer.


CNN based fashion accessory classification model

Employed Convolution neural network with three Conv2D and three MaxPooling2D layer to classify the fashion accessory.


Taxi cab optimal positioning with K-means

Implemented K-Means clustering with 300 maximum iterations to determine the most optimal cab position for Uber cabs.


Stock market prediction of BSE SENSEX using numerical and textual analysis

Implemented five models (Random Forest, Decision Tree, AdaBoost, LightGBM and XGBoost) to compare the capability to forecast the stock market values of SENSEX (S&P BSE SENSEX) with consideration of sentiment analysis. And Random Forest Regressor outperformed other models.


COVID-19 fake/misleading news identification

Implemented NLP for COVID-19 fake news detection. Used tokenization and stopwords for preprocessing, followed by lemmatization for feature engineering followed by scaling with Tf-IDF. Employed several models where logistic regression outperformed other models.


CricFreak

Performed web-scraping, data parsing, pre-processing, and EDA followed by implementing Multi-layer Perceptron Regressor to perform cricket score prediction and angular for the frontend of the web-based application.


US election misinformation identification

Performed NLP misinformation identification for the US election (November 8, 2016). Employed tokenization and stopwords for data cleaning, used word cloud for data visualization followed by feature engineering via lemmatization and scaled with Tf-IDF. Implement four models where logistic regression outperformed SVC, Passive-aggressive classifier and KNneighbours classifier.


Credit card fraud detection

Employed Logistic Regression with random under-sampling to handle the situation of the imbalanced dataset for credit card fraud detection.

Bucks bunny

Developed an android application from scratch on the combined concept of WhatsApp and Splitwise. The idea was forming of groups on the application to split the money on the same. Android studio was used to develop the application


Contact Me

Contact me at

:D

Address

Mumbai

Email

ameyark28@gmail.com