About Me

  • Name:Akshaj Verma
  • Email:akshajverma7 [at] gmail [dot] com
  • Website:www.akshajverma.com
  • Github:theairbend3r
  • Medium:@theairbend3r

Hello There!

I enjoy writing code using R and Python. Into computer vision and deep learning these days.

I worked on time series forecasting and NLP as a data scientist at KPMG for a while. Had a lot of fun developing an R package called 'NNbenchmark' during Google Summer of Code in 2019. Founded a student group called 'Data Science Club' in my university to promote and study applied data science. Love Manchester United, Bob Dylan, and Brooklyn Nine Nine.

If any of the above sounds interesting and you think we would be a potential fit to work together, send me an invitation to connect and let’s talk!

My Resume

  • Work Experience

  • Research Associate

    Indian Institute of Science : Feb, 2020 - Present
    Bangalore, India
    • Computer Vision @ Machine Learning Lab, IISc.

  • Data Scientist

    KPMG India : Aug, 2019 - Jan, 2020
    Bangalore, India
    • Developed a time-series module using Python and Keras to forecast demand for perishable goods to increase profits for a Fortune-500 beverages corporation in the FMCG sector. Experimented with algorithms such as ARIMA, LGBM, LSTM, and 1D-CNN to improve upon the existing accuracy score by 21%.
    • Built a text-classification and sentiment analysis module using Python and Spacy to analyze IT service reviews at KPMG to streamline the pain-points in the service life-cycle. The results from these models were integrated into an overall IT dashboard and used by a team of over 5 analysts.
    • Created a web app using R and Shiny to perform geo-spatial analysis with interactive visualizations on sales visits and extract insights on the operational efficiency for a Fortune-500 beverages corporation.

  • Data Science Intern

    KPMG India : Jan, 2019 - Jun, 2019
    Bangalore, India
    • Developed a web-app using R and Shiny for Entity Resolution based on machine learning and fuzzy matching of entity names and addresses as a proof of concept for a Fortune-500 financial services firm.
    • Built a module using Python, Spacy, and Scikit-Learn to perform text-classification topic-modelling, and sentiment analysis on code-mixed 'Hinglish' customer reviews for a Fortune-500 automotive company.

  • Software Developer

    Google Summer of Code : May, 2019 - Aug, 2019
    India
    • Selected for the Google Summer of Code for the The R Project for Statistical Computing where only 17% of the applicants were selected from all over the world.
    • Taught myself R Package development to build 'NNbenchmark', which provides methods to validate regression based neural network packages that use 1st and 2nd order optimization methods on CRAN.
    • Wrote a meta R script to automate the benchmarking pipeline for 49 different packages.
    • Designed experiments and created R markdown notebooks for ensure reproducibility of results.
    • Developed a responsive website. | akshajverma.com/NNbenchmarkWeb

  • Founder and President

    PyData Manipal (previously Data Science Club) : Aug, 2018 - Aug, 2019
    Manipal, India
    • Founded Data Science Club to democratize machine learning knowledge among the student community.
    • Streamlined strategies for marketing of the events and workshops conducted by group using statistics with R.
    • Recruited and trained 12 team members of varying skill levels in the field of data science, deep learning, and web development.
    • Successfully conducted and organized workshops on Deep Learning for over 100 students.
    • Led 3 sub-teams within DSC Manipal create 3 different deploy-able machine learning projects over a period of 6 months.
    • Developed a responsive website. | dscmanipal.com

  • Telemetry Team Lead

    SolarMobil Manipal : Dec, 2017 - Jul, 2018
    Manipal, India
    • Worked in a multidisciplinary team of 40 people to launch SM-S2, India’s first 2-seater Solar Electric race car.
    • Experimented with Solar Irradiance prediction using deep learning with python.
    • Implemented real time tracking of the vehicle using a GPS module, Raspberry Pi, and Google Maps API.
    • Integrated various sensors, motors, solar-panels, and a battery-management-system together with a CAN bus using microcontrollers.
    • Recruited and trained team 5 members for the Telemetry team.
    • Developed a new responsive website for SolarMobil. | solarmobilmanipal.in

  • SCADA Engineer

    SolarMobil Manipal : Aug, 2017 - Dec, 2017
    Manipal, India
    • Worked in a multidisciplinary team of 40 people to launch SM-S1, India’s first 4-seater Solar Electric Vehicle.
    • Taught myself Arduino and the python programming language to configure and obtain readings from different sensors.
    • Calibrated and interfaced different sensors like current sensor, speed sensor, thermistor, ultrasonic sensor, and gyroscope with Arduino/Atmega microcontrollers.

  • Summer Engineering Intern

    Reliance Jio : Jun, 2017
    Navi Mumbai, India
    • Documented research on LTE networks ranging from the placement of the eNodeBs to the nitty-gritty of EPC(Evolved Packet Core)and E-UTRAN to understand the actual RJIL network setup in Punjab, India.
    • Proposed 10+ cell tower placements to improve cellular reception within the RJIL network setup in Punjab, India.
    • Mastered internal tools used at JiO such as JCP, a big data analytics platform.

  • Education

  • Bachelors Degree

    Manipal Institute of Technology : 2015 - 2019
    Manipal, India
    • Bachelor of Technology.
    • Major in Electronics and Communication Engineering.
    • Minor in Data Science.
    • Awarded the leadership award for outstanding achievement in areas of leadership and consistent demonstration of all-round leadership potential.
    • Electronics engineer at a university funded solar race-car team.
    • Founded a data science group to democratize machine learning knowledge.

My Skills

Machine Learning

PyTorch, Keras, Scikit-Learn, OpenCV, Tidymodels

Data Science

Matpltotlib, Pandas, Seaborn, Dplyr, Ggplot, Tidyr, Shiny, Git/GitHub

Web Development

Flask, Django, SQL, Bootstrap, Beautiful Soup, HTML, CSS

Communication

Vital business value materializes by effective communication of ml models.

Teamwork

If you want to go fast, go alone. If you want to go far, go together.

Business Awareness

Outcomes and value make more sense than process and complexity.

Tools and Languages

Python80%
R90%
SQL75%
HTML5/CSS70%
PyTorch70%
Flask/Django75%
Shiny85%
GoogleFu99%

Projects

Neural Network Pruning

Prune neural networks to reduce parameters and increase CPU inference time using PyTorch.

NNbenchmark

An R package to benchmark neural-network packages on CRAN as a part of GSoC'19.

How to train your neural net

This repo contains deep learning models on various NLP, Computer Vision, and time series forcasting tasks using using Deep Learning with PyTorch.

geoScope

geoScope is a web app built using Flask and OpenCV that uses deep learning for natural disaster management.

Deep Neural Network from Scratch | R

Deep neural network made from scratch using R.

Schools In Need

Increase diversity in Specialized High Schools in New York using data driven methods.

Kaggle User Survey 2019 - Visualization

A dashboard made using R, Flexdashboard, and the Highcharter library to analyze Kaggle’s user survey conducted in 2019.

A Data-Driven Story of Airbnb

Analyze the data, identify problems and opportunities, and come up with insights to increase revenue.

GgPlot ‘Em All | Pokemon

Analysis using data visualization of the Pokemon dataset.

Solar Irradiance Prediction

Time series modelling of solar data using R. A web app is developed using Shiny.

Server log analysis

Analysis of logs generated by web servers using Deep learning.

Neural Network from Scratch | R

Single layer neural network made from scratch using R.

R Pokemon Legendary?

Classification of Pokemon as legendary using machine learning.

Awards and Scholarships

  • Awarded the Intel® Edge AI Scholarship to study the Intel® Distribution of OpenVINO toolkit to deploy computer vision applications at the network edge for the IoT on Udacity.

  • Awarded the Bertelsmann Technology Scholarship to study deep learning using PyTorch for in areas of image recognition, sequence and image generation, and deployment of models using AWS on Udacity.

  • Winner of the Bhatt award by the Computer Society of India for best performer at Intech Olympiad'18 for a project titled 'Analysis of logs generated by web-servers using machine learning'. 14 finalists were selected out of more than 20,000 applicants across India.

  • Awarded the leadership award for outstanding achievement in areas of leadership and consistent demonstration of all-round leadership potential by Dr. Srikanth Rao D, the Director of Manipal Institute of Technology, Manipal.

Get in touch!

Connect with me on LinkedIn or send me an email.

Connect