UK based customer wanted to login into various recruitment web portals and download all job listings for consolidating and downstream processing – where profiles are matched to open jobs and recommended applicants are generated.

For scraping the Web portals, we’ve used the power of Selenium Web Driver and HTMLUnit to scrap and read the data from various portals using class ids, x-path. The Selenium web driver executes on Client machine or standalone server and HTMLUnit uses virtual JVM to execute and grab the data at a scheduled interval. Further, we’ve used Apache SOLR to index the scrapped data and make it structured.

• The Selenium Driver executes using a predefined Web Portal properties URL, Credentials, etc) to scrap the data.
• The properties are configurable from Admin Portal to get the data using class id, x-path.
• Configurable content search from the Web Portals with the help of modifiable data dictionaries.
• We have used Apache SOLR for indexing and formatting the data.
• We have used multi-threading that helps for faster response and concurrent operations.

India based Financial Services customer wanted to automatically extract data from uploaded PAN card and verify it.

Here we’ve used various OCR Tools to read and scrap objects from various image formats like jpg, png, tiff, etc.

• Google Vision API
• Asprise OCR
• ABBYY OCR & NLP

• Load PAN Image in memory.
• Convert Image to bytes.
• Pass Image bytes to google vision API.
• In return, it give all text on the image.
• Parse that text according to standard PAN format using string tokenizer.
• Get different information i.e. Name, DOB and PAN number from that.

India based Financial Services customer wanted to automatically extract real-time market data and news from target websites for use in customer facing investment portals.

The important part here to authenticate and login and then read the real time market data and process it through different channels. The steps are defined herein:

• GET /loginPage for example https://www.abcxyz.com/login?goto=market
• Select the first username tag
• Select the first tag
• Set the value attribute for both inputs (username and password)
• Select the enclosing form, and submit it.
• Check the logout link presence to verify we're logged in.
• Prints the cookie to the console
• Go to market page (https://news.abcxyz.com/market)
• Check table id and table class.
• Robo is going to select the tags that have a class `itemlist`.
• Then Robo will iterate through this table, and for each tr and td select the news headline with link, comments data and no of points to that news.
• Robo will put it in JSON, using Jackson library, to map all data in JSON format and push it to DB for further processing
• Finally, Robo will find out the logout link and click on it to log out.

India based Financial Services customer wanted to automate customer onboarding process for Wealth management.

• All the key steps that required manual verification of uploaded documents for KYC were automated
• OCR was used to extract data from uploaded documents: ID Proof, Address Proof, Bank Statement
• OCR was used to extract data from uploaded Bank Cheque Image
• OCR was used to match uploaded or entered signature with the Signature extracted from PAN
• Vision was used to match live photo taken on the app with the image extracted from PAN or Aadhar card
• E-Sign was used to get user consent and automate the generation of the agreement and customer application form
• Amazon Textract and Rekognition were extensively used to deliver the RPA solution

India based customer wanted to bulk process invoices received over email from vendors and automate the upload of the invoice data into backend ERP systems for Accounts Payable.

• Java based email reader using Camel Framework was implemented
• Extracted invoices were uploaded to FTP folder and onto AWS S3 for storage
• OCR was used to extract data from uploaded invoices into JSON format and stored in PostgresSQL database
• ERP APIs were called and the JSON payload passed for processing