Sei sulla pagina 1di 6

Shubham Mankodiya

Data Scientist

SUMMARY
 3+ years experienced, meticulous & result-oriented Data Science Expert armed with an
analytical acumen in econometric modelling, algorithm development & machine learning
methodologies.
 Possesses a proven track record of setting up the data science function for a leading
hospitality firm, in addition to rendering consultancy services for a Fortune 500
company.
 Proficient in deploying multiple algorithms and techniques such as the k-NN algorithm,
multivariate regression, sentiment analysis, etc. to create products that deliver a direct
impact on the bottom line of organizations.
 Worked on multiple projects of BFSI, Health Care, Retail and Automotive domain.
 Data Driven Problem Solving
 Having strong knowledge in Statistics, Feature Engineering, Model Selection, Model
Evaluation, Feature Scaling to build accurate machine learning.

Work Profile
Analytics & Machine Learning Methodologies

 Conducted extensive research on revenue management and pricing analytics in the


hospitality sector

 Applied various machine learning techniques to build dynamic pricing models and
maximize profits

 Gathered pricing data from different aggregators by performing web scraping in


Python for competitive analysis

Optimization & Algorithm Development

 Developed an algorithm for yield management using the concept of price elasticity of demand

 Deployed multiple loss minimization & optimization techniques

 Led the development of a hotel performance assessment and pricing analysis platform created
via k-NN Algorithm

 Created a recommendation engine to suggest an ideal cluster price for various identified hotel
segments

Statistical Modelling & Analysis


Shubham Mankodiya
Data Scientist

 Created multivariate regression-based attribution models using ad stock analysis


from digital marketing data

 Developed segmentation models using K-means Clustering for exploring new user
segments

 Predicted a customer’s likelihood to book hotels at a given point in time based on the
booking points

 Conceptualized and implemented a sentiment analysis tool to rate hotels based on


subjective customer reviews

Regression Modelling
 Directed model development, validation, testing and implementation of analytical
products and applications
 Developed an additive scoring model for QSM & a logistic regression model to yield a
K-S statistic of 51.5
 Tested and implemented decision trees, random forests and ensemble models via
bagging and boosting

Data Management & Data Mining


 Deployed advanced text mining algorithms to identify search intent latent in
individual keywords
 Employed Principle Component Analysis to analyze collinearity and reduce the
dimensionality of datasets
 Applied Bayesian Model Averaging (BMA) to combine individual keyword-level models

Forecasting & Functional Enhancements


 Analysed internal/external data sources to identify factors impacting change in price
of stocks & forecast volatility
 Led functional enhancements on Reporting & Analytics by deploying R embedded
with C# & ASP.NET
 Integrated data from Oracle ERP & point-of-sale systems to drive strategic planning
& forecasting

Predictive & Statistical Modelling


 Deploying complex predictive models like neural networks, time-series analysis &
simulation to forecast sales
 Created statistical models (ML algorithms/Regressions) & cluster groups (Kohonen
maps/k-Means) to boost market share
 Independently performed visualisation analysis of the assigned database with
Tableau, pivot tables in Excel, etc.
Shubham Mankodiya
Data Scientist

Technical Skills
Tools : Python, R , PostgreSQL, AWS, MongoDB,
MapReduce, Spark, Linux
Packages : Scikit-Learn, Numpy, Scipy, Pandas, NLTK,
BeautifulSoup, Matplotlib, Statsmodels,
Jupyter Notebooks, Tensorflow , Keras
Statistics/Machine Learning : Statistical Analysis, Linear/Logistic Regression, SVM,
PCA, Ensemble Trees, Random Forests, Clustering,
Graph Theory, Recommenders, Regularisations
Cloud : Azure Machine Learning , Google Cloud , IBM Watson

Container Platform : Docker (kubernetes)

KEY PROJECTS:
Driver Score Algorithm for Pay as You Drive model and
Usage based insurance
K-Nearest Neighbour , SQL , Numpy

Applying machine learning solution on automotive datasets. The Data come from the car’s
attached peripheral device called OBD-II.

In this project , client has provided data sets which contained the users trip data which
consists of Cars dashboard data , weather data as well as maps data.Project required to
read bulk data from network path defined in configuration. We have developed a scheduler
which runs every hour and read files from network, these files are passed in model to pre-
dict the output. Output are saved in two formats, one in flat file and second is in database,
which can be used to generate reports.

Challenge was to develop an Algorithm that can provide a driver score to insurance com-
pany to identify the driver is Driving safe or not. We have applied regression and clustering
techniques for achieving best results. As a result we have developed technique that can
identify the safe drive and unsafe drive .

Responsibilities:

 Requirement gathering from the Business Team.


 Data Analysis
 Developing Hypothesis and techniques for best fit score
 Client Communication
Shubham Mankodiya
Data Scientist

 Validation of Driver scores


 Deployment
 Optimization & Algorithm Development
 Weekly meeting.
 Working with Clients team for Deployment and Testing

Artificial Intelligent Chatbot


IBM Watson
Chat-bot for Gynecologic disease identification Watson Analytics , Basic details are
provided in a chart by client , from chart our responsibility to convert it in a chat-
bot sequence and applying machine learning techniques on it. Developing Intent ,
Entities and dialogues for given data. Data manipulation user inputs and testing.

Responsibilities:
 Gathering data and requirements
 Architecture for Continuous integration
 Model validation/Deployment.
 API integration with Python

Text mining
NLTK, Spacy, Flask

Project was aimed to maximum use of NLP for Legal Domain .


The Legal firm wanted their some of cases wants to be summarized in a certain
way that they can get the idea of cases easily , we have used advanced text min-
ing algorithms to identify search intent latent in individual Keywords, Employed
Principle Component Analysis to analyze collinearity and reduce the dimensionality
of data-sets.Bayesian Model Averaging (BMA) to combine individual keyword-level
models.

Responsibilities:

 Requirement gathering from the Business Team.


Shubham Mankodiya
Data Scientist

 Stem and lemmatize texts.


 Text Cleaning and Feature Generation.
 Implemented Ranking Function.
 Bayesian Model Averaging
 Model validation/Deployment.
 API development using Flask.
 Weekly client meeting.

Recommendation
Collaborative Filtering

Real estate firm wants property recommendation to dealers. Client has large num-
ber of property data and wanted recommendation system to their dealers. Data is
sparse, so It suffer from scalability issue which is solved using cluster approach
and new listed property doesn’t get much attention which is cold start problem
and solved by designing hybrid similarity approach.

Responsibilities:

 Requirement gathering from the Business Team.


 Matrix Generation from raw data.
 Identify and solved cold-start problem
 Solved sparse data issue using clustering approach.
 Implemented and Designed different similarity algorithm and recommendation.
 Model validation/Deployment.

Pricing Rule Engine


Pentaho , Elastic Search , Logtash , Kibana

Automotive Company Parts data , Automotive company wants to sell their car
parts for user specifically , by car age , milage , travel time and looking at the oth-
er factors . We have used PMML (Predictive Modeling markup language ) for devel-
oping machine learning solution with Pentaho.

We have used ELK (Elastic Search , Log Tash, Kibanna) stack for Real time data
analysis , full text search , for routing log data and visualizing log data.

Responsibilities:
Shubham Mankodiya
Data Scientist

 Requirement gathering from the Business Team.


 Data Understanding and Architecture Understanding
 Identify and solved cold-start problem
 Developing ELK Stack
 PMML file for Model
 Model validation/Deployment.

OCR - Handwritten Lines Detection


Opencv ,Python , Tensorflow

Client wanted to convert their library books into digital format to save it for
future.The challenge was to identify correct writings with a high accuracy of nearly
99%. As a projectile, we need more images to form the deep learning model
generated by the image enhancement technique. In a second step, read the image
using Open CV and apply the learning transfer to develop the model.As a result,
our image classification model obtained a satisfactory default image classification
with a Type 2 error of zero.

Responsibilities:

 Collection of requirements of the Business team.


 Increase training data using the image augmentation technique.
 Used the concept of transfer learning to obtain good accuracy.
 Implementation of the Deep Learning Image Classification Model.
 Model validated in thousands of images.
 Implementation of the model in GPU.
 Weekly Scrum meeting.

Education:
Master of Computer Application , Business Analytics as a Major subject
Bachelor of Computer Application

Potrebbero piacerti anche