URL shortening service shortens any given URL to a custom size based on the domain name of the shortening service. Created the APIs using Rust, Tokio web, Diesel.rs and MySQL.

Designed and developed the algorithm for creating the bloom filter. Used a test catalog for data which was stored in MySQL database. Integrated this with the Role based access control mechanism.

Determining Users' access to various digital artifacts is a canonical problem in a large scale enterprise system.

Developing a consistent role-based access control as a service with consistency, scale and, performance in mind. The application supports managing various role-based hierarchies with support for grouping and aggregation of resources.

Writing the service primarily in Rust Programming Language.

The algorithm is used for multi-agent path planning. The algorithm is much faster due to its asynchronous nature. Due to its decentralized and state aware nature, it can be used in a wide variety of environments. This empowers complete autonomy for several AI-based robots.'

Implemented and demonstrated the algorithm in Java.

Implemented the Support Vector Machine Algorithm in Python. Aggregated and cleaned the data for the application. Integrated the algorithms with ElasticSearch, Logstash and, Kibanna for real-time monitoring. Achieved an Accuracy of 96%.

Won the runner-up prize in PICT's Impetus and Concepts 2017 competition.

Implemented in Python

Led a team of 4 to implement various Machine Learning Algorithms in Python such as the Support Vector Machine, Entropy-based Discretization and Fuzzy C-Means clustering for real-time DDoS detection in Software Defined Network. Achieved an accuracy of up to 94%. Won the 2nd prize in the Open-source category in a national level competition .

Implemented the Interruptable Multi-agent Asynchronous Decentralized Space-Time cooperative algorithm for path-planning in Python. Achieved a speed-up of 60% over the existing solution.

Designed and Implemented the Parallelized Support Vector Machine Algorithm by implementing the parallel interior-point method to optimize the training phase. Achieved a speed up of upto 24% whilst keeping the memory footprint low.

Implemented the Bloom Filter Algorithm for a recommendation system which recommended items from a catalog to the user. Achieved a speed-up of 2x than the existing solution by also reducing database lookups.