We like solving interesting problems through technology. In this pursuit, we built some interesting solutions. Here are some at a glance.
Automated Photo Moderation using Amazon Rekognition
A large online matrimony company receives close to 20,000 profile uploads a day. They had a team of 12 to moderate these images and approve them for listing. This increased the time to listing and had manual errors in moderation.
Powerupcloud built an API based automated moderation engine using Amazon Rekognition service. While the uploaded images are fed to Amazon Rekognition for image analysis, custom algorithms were used to classify the images under 9 parameters like nudity detection, celebrity detection, blurred images etc.
Private Dropbox on Amazon S3 with Secure File Sharing
One of India’s largest insurance broking company sends several documents to their partners as emails on a daily basis. The security recommendations and heavy nature of the documents demands a Dropbox type platform to share them. The customer was looking for a private Dropbox to meet this requirement.
Powerupcloud built a private Dropbox platform on AWS in under 15 days. This platform allows the recipient to login using an AD authenticated id and download the file. The files are enabled with signed keys and hence can be downloaded only once.
Attrition Prediction using Azure Machine Learning
A large ITES company wanted to build a prediction engine to predict the attrition rate and identify employees who are more likely to leave in next 3 months. This will help them reach out to the employees proactively and address their concerns thus bringing down the attrition rate and save on hiring costs.
Powerupcloud used history data of all employees who left the ITES company in the last one year as the sample set. The prediction models were built on Azure Machine Learning platform. The engine predicted that 69% might leave in a sample set of 1000 employees and the actual figures were very close at 74%.
eCommerce Recommendation Engine using Spark
Google was looking for smart recommendation engines for their online festivals - Great Online Homes Festival and Grand Diwali Mela. In addition to recommending products to customers, the requirement was also to build a prediction engine to predict the time spent by the visitor in the web page so that the recommendations can be pushed before the visitor leaves the page.
Powerupcloud built a real-time product recommendation engine and time predictor engine using Spark to push the right recommendations just before the user decides to leave the page.
ELK Stack for Massive Log Analysis
One of Powerupcloud’s customer was looking to implement an enterprise grade log management engine like Sumo Logic or Splunk to analyse the massive amount of logs (100 GB/day) generated by their system.
Powerupcloud implemented ELK stack for the customer helping them organize all their system and applications logs at one repository. Powerupcloud also built custom rule-based alerting system to generate alerts based on continuous log monitoring. This helped in massive cost savings and brought in great flexibility in writing and managing custom rules.