Green leaf

Pankaj P - Python developer
Profile last updated:Aug 10, 2020
Last activity:Sept. 1, 2021, 12:04 a.m. UTC

Pankaj P

I am ambitious and driven. I am always looking to expand my knowledge and grasp any opportunity to learn something new. I love what I do and hence I have a steady source of motivation that drives me to do my best.

Skills:   Python  ·  Machine learning  ·  SQLite  ·  Numpy  ·  Jupyter  ·  Deep learning  ·  keras  · 
Weekly Availability: 32 hours
Sun 05 - 11 22 - 04
Mon 05 - 10 22 - 03
Tues 05 - 10 22 - 03
Thurs 05 - 10 22 - 03
Fri 05 - 10 22 - 03
Sat 05 - 11 22 - 04

Invite for Interview
Hourly Rate: $20.00
Experience: 2 yrs
Python: 1 yrs
Machine learning: 1 yrs
Engineer's Devices:
Android Tablet


Vellore Institute of Technology     Test Classification and Summarization With Amazon Review     Software Developer     Course Project
Dec 2019 - May 2020

  • The procedure adapted was to firstly clean the reviews (remove stop words, expressions) using functions built into the NLTK library, which perform tokenization and vectorization.
  • Subsequently I prepared the neural network model and fit the data. Lastly I performed the training and evaluation. Naïve Bayes and Linear SVM were utilized for text classification purposes.
  • The code was executed on jupyter notebook. Deep learning neural networks along with Machine learnings were conducive to the project

Skills used: Deep learning, Jupyter, Machine learning

Link to the github:
Vellore Institute of Technology     Image super resolution using Autoencoders     Software Developer     Passion Project
Jan 2020 - Mar 2020

  • An auto encode in essence is a network wherein the original input is first encoded, subsequently the compressed representation is then put through a decoder to get the reconstructed input.
  • The encoder and decoder consisted of 5 layers each, in order to achieve the transformation from low resolution to high resolution using deep learning models.
  • The code was executed on Jupyter notebook, and the neural networks were constructed using keras and tensorflow.

Skills used: Deep learning, Jupyter, keras

Link to the github:
Vellore Institute of Technology     Irregular Human Activity Behaviour     Software Developer     Course Project
Jul 2019 - Dec 2019

  • I trained a movement Impact Map on a training dataset which consisted for videos where no irregularities were observed.
  • The purpose of a Movement Impact Map is to predict the future behavior of people based the data provided. During testing, the machine learning model predicted the future movement of the people.
  • If there was any discursion from the predicted movement, it flagged the activity as irregular. The machine learning model was trained .
  • The code was written in python.

Skills used: Python, Machine learning

Link to the github:
Vellore Institute of Technology     Virtual Museum     Software Developer     Passion Project
Jul 2018 - Dec 2018

  • In order to build this digital museum, the application design and user interfaces were built using tkinter library of python.
  • For the backend I used SQLite to store the images and video files(if necessary).
  • The images were converted to a numerical representation in order to ease the process of storage (the conversion was achieved using numpy library).

Skills used: Python, SQLite, Numpy

Link to the github:


Vellore Institute of Technology    B. Tech  (Computer Science)
Jul 2017 - Jul 2021


Applied Machine Learning in Python    Coursera-University of Michigan
Mar 2020 - May 2020
Skills learnt: Plotly, Numpy, Pandas, Matplotlib, Machine learning
Applied Text Mining in Python    Coursera-University of Michigan
Apr 2020 - Jun 2020
Skills learnt: Google NLP API, Deep learning