Data Scientist | NLP, Sentiment Analysis &Text Summarization

About the Company:
Amicus Technologies™ is a data analytics startup, founded to solve the problem of consumer complaints in the market of High Value Purchases. It solves the problem by entering at both the pre-purchase and post-purchase stage. It is the only startup in the world which has figured out a way to monetize the huge amount of data languishing all over the web in the form of consumer complaints. Amicus Technologies has built two key products- Amicus Shopping Assistant (AmicShop- Pre-Purchase) and a Complaint Amplification Mobile App- Amicus Consumer Voice (AmicVoice- Post-Purchase).
Pre-Purchase via the Amicus Shopping assistant, Amicus reduces the likelihood of a consumer complaint by showing the user the RealCost of a product && seller. The RealCost is curated by AmicRank algorithm by looking at 40+ variables from 3-5 Lakh data points collected from social media, ecommerce websites and complaint forums. The AmicRank algorithm analyzes 40+ data points by organizing them into Risks – likelihood of something going wrong, Rights – the rights when something goes wrong and Response – the response when you invoke your rights.
Post-Purchase via the Amicus Consumer Voice, Amicus increases the likelihood of a resolution by allowing the user to create a complaint without typing a single word. Our engine then verifies the legitimacy of the complaint. We have reverse engineered sentiment analysis tools to stori-fy complaints using keywords. This increases the chances of the complaint being noticed by the brand. We also increase the reach of the complaint, by allowing the user to post it on multiple websites and social media forums, in a single click.

Website: http://amicshop.com/

Desired Experience:  0 – 2 years

Salary: INR 3 - 8 LPA

Tentative date of Interview: will be communicated post registration window

Job Description:
-This role requires you to build upon the summarization engine which distills more than 5 million reviews (for smartphones) to give you 8 key facts and figures, per smartphone, brand and seller.
-You can read the methodology currently followed by us here: http://www.amicshop.com/methodology.
-We are currently at 75% accuracy with summarizing reviews and complaints using Sci-Kit, with a goal to reach 95%.
-This is an extremely challenging role which requires us to deal with unstructured data and build training sets which have english as a second language.
-This role also requires summarization of structured text, i.e. Terms and Conditions, using semantic analysis tools, in real-time.
 
Skill Set Required:  
-Knowledge of NLTK and ML rules and libraries.
-Past experience in using Regex, Pandas and Sci-Kit.
-Experience in implementing Sentiment Analysis tools.
-Ability to ask to lead analyst teams which will build the training sets.
-Experience in data collection and optimization of input funnel to improve training sets.
-Experience in data visualization on the Front-end.
-Build bag of words using POS tagging, and unsupervised learning to manually eliminate double negatives, false positives, and false negatives.

Interview Process:
-2 Technical Round
-HR Round