Jul 2024 – Present · 1 yr 11 mos

Jul 2021 – Jun 2024 · 2 yrs 11 mos

• This project is a content management and research application by WNS, first implemented for client. It offers features like URL content scraping, text cleaning, and using GPT for text transformation, summarization, and media post creation. It also includes court case summarization through Azure Document Intelligence and Langchain library, translation services via Azure, and internet research capabilities using Google or Bing, with content formatted by GPT. Fitch primarily uses InfoTurf's research feature to gather and format information about infrastructure projects.
• ZTA implementation using Instasafe
• SSO implementation - POC on Okta, AWS cognito, google auth etc, SAML federation, IDP, social media login, authentication and authorisation
• Scalability enhancement- Optimised API which can serve data 200000 rows in one request, by using this API we can get 1M records saved in 10 mins using REST API.
• Scalability enhancement - System enhance to 1500%, previously the system can process up-to 10000 keywords and now it can process 1,50,000 keywords.
• Assets ->This project works on ETL methodology, Extraction is done on crawling layer where data is picked
• for assets from different sources and gets stored in MongoDB, after extraction data moved to transform
• layer where data gets stored monthly and portal is another system, Portal is a front-end tool for the client where
• they need to view the data of different Assets, and perform sanity on the
• data being accumulated by their crawling team. Through this portal a user can schedule a crawler and collect
• data for the selected sources.

Mar 2017 – Jun 2021 · 4 yrs 3 mos

• The Chopras -->> This project is a front-end tool for the client where they need to view the data of different Universities
• offering different programs and courses all over the world, and perform sanity on the data being accumulated
• by their crawling team. After the crawler is executed successfully the data gets collected in the database, and he
• can then view it in a separate module in the portal, a user can login to the system as ‘Uploader’, ‘Approver’,
• ‘Admin’, ‘Super Admin’ and ‘View Only’ roles. An uploader can perform first level verification of data and send
• the data for Approver’s approval. If the data gets approved by the Approver it is ready for being displayed on
• the Home Screen, from here this data can be downloaded as en excel. We are maintaining 3 levels of data
• record for a single University Program i.e. n-2. So that a user can view the changes made in a specific entry by
• a University. The crawling part is being taken care of by another team, which has exposed three APIs for us to
• get the status of a scheduler, accept a new scheduler request and as soon as a scheduler is completed pull the
• data from the crawling engine and save the data in our database.
• Smart News -->> Extracting the news from different web sources and Microsoft Bing API based on keywords and their
• related aliases, crawling different news sites for getting the news. Enter all of the collected news articles in the
• database and removing duplicity in the data at db level, performing analysis of the text and giving it a sentiment
• score on the basis of an algorithm, where the negative data decides the negativity score of the news article,
• at last dispatch the news as per client to different users and stakeholders via email, as an attachment. These
• news are also posted to different portals which are used internally, by an automated script. Keywords and their
• aliases are handy and can be entered in the database at any point of time, same goes with the client and user
• mapping.

Jan 2016 – Mar 2017 · 1 yr 2 mos