Dotodo is a distributed, large-scale, low latency platform that seeks to connect people wanting to do similar things.
Atlogys CTO's helped shape the product by providing the inputs.
Atlogys CTO's engineered Dotodo to be a Railes based nosql system running on the amazon cloud. Some of the complexities we helped design, engineer and test into the system are as follows:
Allows us to automatically detect todo's written by different users that may have the same intention. E.g. We should match the following:
This involves a mix of manual and automatic effort around intelligent semantic detection, synonym identification and simple text based approaches. Travel tree based algorithms are used to provide results of the highest accuracy.
Activities on the site are posted to fb and tw using their social API's The challenging part is that we back-fetch activities on fb and tw related to our events and show them on the dotodo site also. Challenges involve rate limiting, throttling, data normalization and bandwidth management.
This is related to the semantic search. Tags are social intentions of people. The goal is to detect and show the most relevant set of tags given a particular set of todos.
Almost 70% of the site is in Ajax. Results are fetched using background processes that are outside the realm of the traditional request/response paradigm. This makes the UI more intuitive and engaging.
The platform is built in the MVC (Model View Controller) framework. The backend exposes a RESTAPI which makes it easy to integrate it with external systems. Whether it means creating a mobile app or an iPad one, the same can be done without any re-haul of the system.
Does not use traditional relational database management database systems. Instead is built on top of document oriented Db systems that are far away from tables and have Key Value pairs for storing data.
The system is deployed and running on EC2.