Content Retrieval System.
Technical Lead (ML)
Developing a system for the K12 domain that retrieves content such as lectures, study materials, practice exercises personalized to user’s requirement/progress in an academic year.
A Real-Time solution for TV Viewership collection and processing for India.
Lead Software Engineer
Television rating point (TRP) index is measured by collecting Viewership information of consumers via population sampling. In general, an IoT solution is deployed across these sampled consumers to collect this information for necessary post-processing (TRP index). But, as India observes a quick up-tick in consumer base and digitization, it demands for a large scale and quick generalization (of samples) solution. In addition to this, considering the infrastructure across India (including remote areas), the solution should cater to challenges such as:
- Nodes robustness towards network/power disruptions. Especially, in handling time, as the Real Time clock (RTC) management can go out of sync leading to misinterpretation of data.
- Node monitoring. Since the data is of critical nature, any tampered (outlier) behaviour/node health issues should be identified immediately for isolating them from analysis.
- Data communication and processing pipelines should handle data that scales linearly (per second) with #nodes.
Addressing such multitude of challenges*, we designed and developed an end-end solution based on event-driven architecture. The solution scales to a million of these Viewership collection nodes and achieves real-time processing of data. Our solution is currently deployed across India (under the Broadcast and Research Council (BARC) body), and is serving as a means for India’s TRP measurement.
HONOR: Received a certificate of appreciation from BARC.
*Please refer to my Resume for detailed information.