Apr 2021 – Mar 2022 · 11 mos

• Designed and implemented Deep Learning pipelines for Document Layout Analysis, Table Detec-
• tion, and Structural Component Extraction.
• Improved preprocessing pipelines with skew correction, noise removal, and rotation handling for
• scanned document images.

Apr 2017 – Apr 2021 · 4 yrs · Pune Area, India

• I am currently working at Wipro Holmes. Holmes is an Artificial Intelligence platform developed by Wipro.
• Built machine learning models for document layout analysis and OCR.
• Developed OCR systems for text extraction and classification using Faster R-CNN, Mask R-CNN,
• and CRNN.

Jun 2016 – Mar 2017 · 9 mos · Pune Area, India

• Aspect Based Sentiment Analysis from text data.
• Aspect Based Sentiment Analysis : At Digitially I have worked on end to end development of
• Aspect Based Sentiment Analysis System. For aspect term recognition and extraction, I designed
• and implemented many novel algorithms. These algorithms are ensemble of supervised
• algorithms like Deep Learning, SVM, Random Forest and unsupervised rule based algorithms.
• Sentiment about extracted aspects were provided by NLP algorithms which uses Naïve Bayes,
• SVM, Random Forest for classification.

Dec 2015 – Jun 2016 · 6 mos · Mumbai

• Bike Recommendation System: Designed recommendation system for suggesting bikes for CredR. This involves analysing features of bikes in which purchaser can be interested and recommending bikes on that basis. This project involves design, development and coding in python and postgresql. This system is Item based recommendation system for bike recommendation.
• Analysing Business data :
• This task involves statistical analysis bike sales data over period of time. The data is processed using machine learning techniques like clustering, PCA. Then this data is visualised and meaningful insights are found which helps in taking the business decisions.
• Accelerometer data analysis and bike state prediction:
• It involves capturing accelerometer data gathered by mobile app. Data consist of accelerometer and gyroscope reading in x, y, z direction Then this data is processed to estimate the state of bike like normal ride, bumpy ride, continuous brakes etc. By applying ML technique the approximate time at which brake, bump occurred are found out . The results from it are used to give recommendation to rider

Aug 2014 – Oct 2015 · 1 yr 2 mos · Chennai Area, India

• Sales Analytics:
• Inspecting opportunity in loss reasons using NLP, Finding out most frequent loss reasons and analyse it on basis of quantities and revenue. Weighting lose reasons. Analysing the association between product sales. Reporting valuable Insights got through Data Analysis.
• Exploratory Analysis:
• It is internal project with aim to reduce data exploration and cleaning time. Designed Data Exploration tools in R which helps to graphically inspect features/attributes in dataset. It helps to summarize data at feature level enable studying its statistical properties like mean, mode, median, most frequent observations. This package also uses different techniques for outlier detection and missing value imputation.
• Business Intelligence:
• This project involves analysing sales and purchase data across different merchants, clients using Teradata SQL. The main tasks in this project include Categorising merchants on basis of different products sales, volume of sale, geography of sale, quantity of product sale. Searching for the measure changes in pattern of sale. Inspecting cross border sales/ purchase, Product categories. Understanding products and associated buyers. Inspecting behavior of buyer across all segments.
• Cost Optimisation:
• The aim of this POC was to Identify cost savings opportunities in ordering parts/components through data analytics. Dataset consist of orders of different manufacturing parts required for the networking product. These parts were ordered from different manufactures across different countries from several manufacturers. The main tasks done in these project involve building a forecasting model to predict future cost of parts based on historical data. It also involves Correlation analysis, Cost saving through distributing order to different manufacturer.

Sep 2013 – Aug 2014 · 11 mos · Pune Area, India

• Designing,Developing and testing Data Mining and Machine Learning Algorithms.
• ”Knowledge based elucidation of tertiary structure of protein on their function algorithm
• development and web hosting of prediction server.”
• This project involve designing, developing and testing data mining and machine learning algorithms for various classification and regression datasets. The algorithms developed was Genetic Algorithm, Simulated Annealing, Ant Colony Optimisation Algorithm, GSO algorithm for feature selection and dimensionality reduction of high dimensional datasets. The algorithmic d evelopment for this project was done using Java, Matlab, Octave and R. The measure statistical and data mining tools used in these projects were Weka, Random Forest, SVM.

Jul 2010 – Jun 2011 · 11 mos · Nashik

• TEACHING ENGINEERING SUBJECT LIKE PROGRAMMING LANGUAGES,OPERATING SYSTEM,DATA STRUCTURE, COMPUTER ORGANISATION,COMPUTER NETWORK,DISCRETE MATHEMATICS

Aug 2009 – Jun 2010 · 10 mos · Nashik

• TEACHING ENGINEERING SUBJECT LIKE PROGRAMMING LANGUAGES,OPERATING SYSTEM,DATA STRUCTURE, COMPUTER ORGANISATION,COMPUTER NETWORK,DISCRETE MATHEMATICS