DINESH YELLURI

Created and Automated pipeline and modeling for any type of sensor data to provide Regression, Anomaly Detection and Prediction based on user selection. Created streaming pipelines for Ingestion to Influx(Telemetry/Spectral/Discrete). Created unified streaming pipeline for User Rule Evaluation, Event frame and User Custom Expression evaluation services. Contribute to designing, building, and deploying high-performance production platforms using dockers. Created automated streaming data pipeline to serve 200 models within 10 core CPU power to provide real time predictions at low latency. Created a pipeline to serve 1000+ user created models and make predictions. Created connectors for IOT sources like MQTT, OPCUA, OSIPI etc.

Worked on Re-engineering and Migrating data for Cobol - 600+GB of data containing 32+ diFerent data formats [Experian]. Worked on migrating multiple data sources from old mainframe environment to Hadoop ecosystem [Experian]. Worked on adding validations to new Bigdata based Credit user validation system [Experian]. Worked on creating a Lambda based pipeline for ingestion of data from 250+ greentea tables (Redshift) [Grubhub]. Created Azkaban, Presto, S3 based new ETL pipeline and migrated data for same [Grubhub]. Worked on creating Lambda based pipeline for clickstream events [Grubhub]. Worked on creating and optimizing Grubhub EMR based Cluster Management [Grubhub].

Worked on HBase based SMS/Voicemail storage platform use message conversion to byte array and stored per user with a retention [Comverse]. Worked on merging the Comverse Data Platfrom in Israel to global [Comverse]. Worked on a POC that crawls the web for information about diFerent mobiles and creates/updates the model weekly to predict the price of a second hand/refurbished mobiles and provides best price one can get for the selected mobile for given set of information about mobile [POC].