Local Content and Local Search

* Developed and maintained a deployment system that became a part of the CI/CD pipeline* Developed a metrics server that measured key signals from the deployment system* Maintained a monitoring system that would feedback into the deployment system in case the deployment went bad and auto-rollback to previous good version.Technologies: Java, Python, Mysql, MongoDB

* Building essential platform services shared across multiple components in the backend. * Designing and developing the common-logging and log analytics framework using collector-sink model.* Developed postgis python scripts to load, transform and process map data.* Key contributor to the design of the OSM backend, which includes developing and implementation of adaptor to ingest OSM data and make it available for routing engine, display engine etc.* Maintenance, Performance tuning and releases of Traffic server: a service that provides both real time and historical traffic and incident information.* Developed and integrated traffic feeds. * Developed and implemented historical traffic models for prediction of speeds and incidents.

* Built the indexing of large 2d/3d underwater files for search and play capabilities.* Implemented the classification module for classifying an underwater image into a set of given categories (shale, sand, man-made etc).* Key contributor in developing the batch processor module to improve processing of large amounts of data.

* Maintenance and release of flagship product Vicon iQ.* Wrote the hardware abstraction layer from scratch that connects to cameras and sends tasks like background subtraction, calibration etc.* Implemented the T-Pose Labeler, that correctly benchmarks an object so that it can be tracked.

* Developing scripts to link programs that take as an input a sequence of amino acids and output the 3d predicted structure of the protein.* Clustering methods to analyze biological data.

* Developed a software tool and data to evaluate image retrieval system.* Data analysis using regression - Bayesian methods.* Feature extraction algorithms.

Developed a device to detect faulty speakers and is currently being used in the industry and saving upto 30% of testing time. Supervisor: G. Rajnarayan