Tanmoy patra resume
?
0 likes?63 views
This document is a resume for Tanmoy Patra which summarizes his experience and qualifications. It outlines his work as a data scientist and engineer at Deloitte Consulting where he has led several projects involving pricing recommendations, sales quota optimization, customer dues recovery tools, and other applications to solve problems for major clients. It also provides details of his education including a M.Tech from the Indian Statistical Institute and B.Tech from the Institute of Engineering and Management. Relevant skills and a research internship are also mentioned.
![TANMOY PATRA
Data Scientist & Engineer
? tanmoy.patra009@gmail.com +91-70934-00624 Hyderabad, IN linkedin.com/in/tanmoy-patra-544b7556 github.com/tanmoypatra
INNOVATION
? Hardware Prefetching in Multi-Core
Processors:
Improved performance of Hardware Prefetcher
by increasing the speed–up by 15%, exploiting
different data access patterns and minimizing
prefetch induced interference. Rated most in-
novative by AMD.
? Engagement labeling between customer and
CRM agent on Twitter Data:
Designed a tool to track the engagement and ef-
fectiveness of CRM agents in addressing the is-
sues raised by customer on online social media
such as Twitter, thus enhancing customer satis-
faction.
? Web Crawler for Ef?cient Search Mechanism
on an Educational Domain:
Designed a Web Crawler for an educational in-
stitute to scan through the textual content of
the domain based on a user search to return rel-
evant web pages. Enhanced Scoring mechanism
alongside pre-compute cache made it informa-
tive plus 20x faster.
EDUCATION
M.Tech in Computer Science
Indian Statistical Institute
July 2013 – May 2015 Perc: 78%
[Dissertation] Point of View Identi?cation from
multiple newspapers:
Proposed a novel approach to automatically ex-
tract multiple points of view from different news-
papers on a similar topic to provide users with a
point-of-view wise centric aggregation. This en-
hanced user experience with multiples insights &
opinion in an effortless way.
B.Tech in Computer Science
Institute of Engineering and Management
June 2008 – May 2012 CGPA: 8.82/10
APPLIED METHODS & SKILLS
GLMs Time Series K-mean Clustering
Data Regression KNN Decision Trees
Opinion Mining Support Vector Machines
Random Forests Optimization
R & SparkR Python & PySpark C++ & Java
Excel & Sql Angular JS & Django Gensim
NLP Tableau AWS Lex Dialog Flow
?
Helping clients solve critical problem in
different domains across geographies.
?
Awarded by XRCI for innovation in
Domain Adaptation
EXPERIENCE
Consultant & Data Scientist
Deloitte Consulting USI, Science Based Services
June 2015 – Ongoing Hyderabad, IN
? Pricing Recommendations for one of America’s largest food manufacturer :
Helping client set prices on a weekly basis for a-la-carte and combo items to max-
imize margin for 14K stores. Given the non linear nature of objective function
with several constraints and scenarios makes it challenging. Ongoing engage-
ment currently in beta phase.
? Optimizing the Sales Quota assignment for Sales Rep of a Fortune 50 Global
Technology Firm:
Helped client save over $5 million per ?scal by setting better targets for the Sales
Rep, delivering results to 5 countries - The U.S., UK, France, Germany & Rus-
sia with multiple stakeholders on a time constrained and technically challeng-
ing project. Circulant Auto-Regressive Time Series ensembled with Opportunity
model was designed to adapt to the limited historical data.
? Voice based Customer Dues Recovery tool for one of top US Telecom providers
:
Developing an end to end capability to enable a virtual assistant to handle cus-
tomer calls for payments. Primarily human agents handled the process. Letting
the VA handle the repetitive nature is one of a kind.
? Manufacturing Optimization Model for a major US Tire Manufacturer:
Increase in production of tires by 10% by strategically predicting the feasibility
and schedule of manufacture based on demands. Used Random Forest for rate
models feeded to the optimizer which used heuristics to optimize production ef-
?ciency.
? Designing, building and deploying Algorithmic selling capability to the Sales-
force of a Top 5 US Packaged Food & Beverage Company:
Helped increase the effectiveness of Territory Business Manager’s(TBM) sales
activities by providing them with enhanced sales tools and actionable incremen-
tal sales opportunities.
? Designing and building of uni?ed prototype for unit-linked & non-linked premi-
ums for a Leading Life Insurance Company:
Worked with CRO on client side to design & develop a framework to unify the
?ow of insurance premiums and bonus calculations combining data from differ-
ent sources. Effort saved client 1 weeks every month of 3 full time employees,
with an automated code.
? Commercial Applications for Intelligence (CAFI) :
Deployed a platform worth $2 million to demonstrate due diligence which proac-
tively identi?es individuals and entities that pose a risk in the supply chain by Text
Analytics & Active Learning. A end to end design with Angular js at front end,
Django as backend with Postgresql, Redis as cache with Celery workers.
RESEARCH INTERN
Xerox Research Innovation Centre
Text Analytics
June 2014 – Sept 2014 Bangalore, IN
? Methods and systems to train classi?cation models to classify conversations:
Proposed a technique to adapt machine learning based microblog categorization
algorithm once trained on a domain to apply to a new domain with minimal or
no labelled data from the later. Several adaptation to CRF was made to make it
adaptable to short text format.
PUBLICATIONS
Patent
? Bhatt, H.S., S. Roy, and T. Patra (2017). Methods and systems to train classi?cation
models to classify conversations. US Patent App. 14/872,258. Google Patents.
URL: https://www.google.com/patents/US20170098443.](https://image.slidesharecdn.com/tanmoypatraresume-181031124148/85/Tanmoy-patra-resume-1-320.jpg)
Tanmoy patra resume
- 1. TANMOY PATRA Data Scientist & Engineer ? tanmoy.patra009@gmail.com +91-70934-00624 Hyderabad, IN linkedin.com/in/tanmoy-patra-544b7556 github.com/tanmoypatra INNOVATION ? Hardware Prefetching in Multi-Core Processors: Improved performance of Hardware Prefetcher by increasing the speed–up by 15%, exploiting different data access patterns and minimizing prefetch induced interference. Rated most in- novative by AMD. ? Engagement labeling between customer and CRM agent on Twitter Data: Designed a tool to track the engagement and ef- fectiveness of CRM agents in addressing the is- sues raised by customer on online social media such as Twitter, thus enhancing customer satis- faction. ? Web Crawler for Ef?cient Search Mechanism on an Educational Domain: Designed a Web Crawler for an educational in- stitute to scan through the textual content of the domain based on a user search to return rel- evant web pages. Enhanced Scoring mechanism alongside pre-compute cache made it informa- tive plus 20x faster. EDUCATION M.Tech in Computer Science Indian Statistical Institute July 2013 – May 2015 Perc: 78% [Dissertation] Point of View Identi?cation from multiple newspapers: Proposed a novel approach to automatically ex- tract multiple points of view from different news- papers on a similar topic to provide users with a point-of-view wise centric aggregation. This en- hanced user experience with multiples insights & opinion in an effortless way. B.Tech in Computer Science Institute of Engineering and Management June 2008 – May 2012 CGPA: 8.82/10 APPLIED METHODS & SKILLS GLMs Time Series K-mean Clustering Data Regression KNN Decision Trees Opinion Mining Support Vector Machines Random Forests Optimization R & SparkR Python & PySpark C++ & Java Excel & Sql Angular JS & Django Gensim NLP Tableau AWS Lex Dialog Flow ? Helping clients solve critical problem in different domains across geographies. ? Awarded by XRCI for innovation in Domain Adaptation EXPERIENCE Consultant & Data Scientist Deloitte Consulting USI, Science Based Services June 2015 – Ongoing Hyderabad, IN ? Pricing Recommendations for one of America’s largest food manufacturer : Helping client set prices on a weekly basis for a-la-carte and combo items to max- imize margin for 14K stores. Given the non linear nature of objective function with several constraints and scenarios makes it challenging. Ongoing engage- ment currently in beta phase. ? Optimizing the Sales Quota assignment for Sales Rep of a Fortune 50 Global Technology Firm: Helped client save over $5 million per ?scal by setting better targets for the Sales Rep, delivering results to 5 countries - The U.S., UK, France, Germany & Rus- sia with multiple stakeholders on a time constrained and technically challeng- ing project. Circulant Auto-Regressive Time Series ensembled with Opportunity model was designed to adapt to the limited historical data. ? Voice based Customer Dues Recovery tool for one of top US Telecom providers : Developing an end to end capability to enable a virtual assistant to handle cus- tomer calls for payments. Primarily human agents handled the process. Letting the VA handle the repetitive nature is one of a kind. ? Manufacturing Optimization Model for a major US Tire Manufacturer: Increase in production of tires by 10% by strategically predicting the feasibility and schedule of manufacture based on demands. Used Random Forest for rate models feeded to the optimizer which used heuristics to optimize production ef- ?ciency. ? Designing, building and deploying Algorithmic selling capability to the Sales- force of a Top 5 US Packaged Food & Beverage Company: Helped increase the effectiveness of Territory Business Manager’s(TBM) sales activities by providing them with enhanced sales tools and actionable incremen- tal sales opportunities. ? Designing and building of uni?ed prototype for unit-linked & non-linked premi- ums for a Leading Life Insurance Company: Worked with CRO on client side to design & develop a framework to unify the ?ow of insurance premiums and bonus calculations combining data from differ- ent sources. Effort saved client 1 weeks every month of 3 full time employees, with an automated code. ? Commercial Applications for Intelligence (CAFI) : Deployed a platform worth $2 million to demonstrate due diligence which proac- tively identi?es individuals and entities that pose a risk in the supply chain by Text Analytics & Active Learning. A end to end design with Angular js at front end, Django as backend with Postgresql, Redis as cache with Celery workers. RESEARCH INTERN Xerox Research Innovation Centre Text Analytics June 2014 – Sept 2014 Bangalore, IN ? Methods and systems to train classi?cation models to classify conversations: Proposed a technique to adapt machine learning based microblog categorization algorithm once trained on a domain to apply to a new domain with minimal or no labelled data from the later. Several adaptation to CRF was made to make it adaptable to short text format. PUBLICATIONS Patent ? Bhatt, H.S., S. Roy, and T. Patra (2017). Methods and systems to train classi?cation models to classify conversations. US Patent App. 14/872,258. Google Patents. URL: https://www.google.com/patents/US20170098443.