Challenge: Startup in healthcare- Searching from thousands of books and millions of symptoms, and coming to right remedies to using various algorithms.

Solution: We created our own Search Engine, own Database, Indexing system, that would search through gigabytes of data in fraction of seconds. This was done when ready search engines were not available. This was a smart Lexical Search Engine, with logical operators like AND, OR, NOT, Same para/ chapter etc. with intelligence of context understanding.

The success: This product is currently sold in 150+ countries and is a leading homeopathy expert system. The product is using same engine for their expanded data and working efficiently even after 30 years!

Scenario: A country’s leading Telecom operator, who had their own in house Billing application, written in C. The software had taken care of various parameters, and could create billing plans accordingly.

Problem: In telecom, everyone sells the same services, but you need to package them in different billing plans. One has to keep innovating in this area to create attractive plans. Sometimes the marketing team would come up with some plan ideas, that would not fit into the existing parametric software. This involved a cycle of custom development, testing, parallel run, and then release. This sometimes would lose many months, and the competition could just move faster!

There had to be a solution that would convert ideas to working solution in very short time. There had to be some kind of rule engine!

Options: There were options of junking the current software, and buying and implementing (much tougher than buying) it. Another was to go for readymade rule engine, but this option involved in rewriting the whole code using the Java libraries of this provider. All this looked really heavy.

Solution: A faster, cost effective solution, modern interface and Rule engine working on top of existing software!

We implemented our Rule Engine, creating a simple, custom Object Oriented Rules language that was specifically fit for Telecom admins.

This turned the existing application and database into an Object Oriented one, with the same code and same DB. Now the admins could create a plan hierarchy, inherit from a plan, write new rules In a simple language and launch the plan.

We created our own Compiler and Interpreter for the Rule Engine.

Problem: Island nation - dependent on tourism economy. Also foreigners invested in beach properties/ villas. All this was nice, but problem started when the villa owners started renting the property as weekend gateways. Obviously personal property does not pay commercial tax. So hotels lose business, and government started losing on Tax revenue.

Consultants were suggesting to raid the villas - an intrusive option, that was really dangerous. One wrong raid and the country would have bad name, branding and impact tourism. So how to find the defaulters, without conducting raids?

Observation: This was the key to solving the problem. The villa owners are usually individuals, and not big hotels. They are not likely to have their own marketing department. So how would they market their villas? Of course on property listing websites, yellow pages, classified ads etc. (Beach villa available for weekend, with pool etc)

Solution: After connecting the observation and the problem - solution was obvious. (Though technologically not simple) We created a web crawler to crawl through hundreds of listing websites, find relevant properties, check them with the commercially registered property database, and if not found, send this information to the Inspection department. This worked, and the problem was solved.

Challenge: A leading carpet exporter was exporting to buyers who would sell as their own brand. The client wanted his own brand to reach world wide directly. But typical e-commerce model would not work here.

This involved creating an innovative business model with supply chain framework, for a local business to go in the international market, and create E-Commerce model. Challenge was, Carpets just can’t be sold online, as the touch and feel, are the most important aspects. Also colors look different on each screen and may look different in real life. If this was to be distributed to retailers - the challenge was, that how would we form distribution – why would a retailer stock our products, and how would he manage return of unsold inventory?

Solution We solved these problems by

• creating a business model, where online business does not bypass retailer, but partners them.
• Retailers don't need to keep large inventory.
• People get true colour and feel of the carpet.
• Application that allows people to see the carpet design with different colour combinations, by allowing them to customize online.

Most optimal way for doing pool testing based on the probabilistic model, and with moving concentration of positive case density. Our founder Parag, won the PAN IIT challenge for best algorithm.