Siva Saravanan

• Enhanced and optimized the AirwaveSP RAN analytics platform's multi-source data ingestion process; significantly improved the performance of the primary MemSQL ingestion stored procedure, reducing its average execution time by approximately 83% through logic changes
• Developed comprehensive Grafana dashboards to monitor data pipeline performance (throughput, latency) for the AirwaveSP application
• Contributed significantly to the SPACE program by re-architecting data pipelines for high-volume RAN/CORE network monitoring data, migrating the RTT application from Druid/Spring to a modern Flink, Kafka, and Cassandra-based platform.
• Led the complex data migration of the UCS Eclipse portal from Druid to Cassandra (NoSQL), designing the new data architecture and achieving $1M in cost savings.
• Developed data ingestion pipeline modules using Apache Flink for the ANPD dataset, processing Parquet files from S3 via Kafka
• Engineered performance optimization PoCs for data pipelines, developing Flink aggregation jobs (e.g., 1-min granularity) and resolving complex Avro serialization issues in Kafka data streams
• Designed and built efficient data retrieval APIs (Node.js/DuckDB) and high-throughput asynchronous services for large-volume Cassandra data reads
• Analyzed complex existing data flow architectures involving Apache Spark and MemSQL (used as RAN analytics data warehouse for AWSP application)
• Initiated and led technical exploration of Starrocks database, researching setup, S3 integration, and query acceleration techniques (colocate/lateral joins), presenting findings to the team
• Developed robust data ingestion, aggregation, and query services utilizing Cassandra and PostgreSQL.
• Mentored junior engineers on data engineering concepts and project tasks.
• Led migration of critical legacy applications (Canvas eApps) to cloud storage (AWS S3), establishing CI/CD build pipelines for deployment.

EDGEAI Platform & Netflow Processing:

• Developed and optimized Apache Spark applications to consume high-volume Netflow data streams from Kafka, perform metadata tagging, and ingest into Druid databases using Avro format for the EDGEAI analytics platform.
• Drove significant performance improvements in Netflow Spark jobs by correcting compression settings (LZ4), standardizing on Avro format across applications, eliminating data duplication sources, and optimizing logging configurations.
• Implemented Spark ML (K-Means clustering) for outlier detection in MVD antenna datasets, contributing to improved FWA service quality analysis within EDGEAI.
• Evaluated Redis integration with Spark/Kafka for data handling, providing key insights into design constraints for streaming data consumption.

Apollo - Graph DB & ML:

• Explored and evaluated graph databases (Neo4j, AgensGraph) for the Apollo NextGen Processing Engine PoC.
• Developed microservices to process and load network topology and alarm data into Neo4j.
• Implemented a Neo4j trigger for automated alarm detail collection and push to messaging systems (VMB).
• Developed and demonstrated a Proof-of-Concept machine learning model (scikit-learn) for root cause alarm prediction.

Data Migration & Performance:

• Engineered Spring Boot microservices using Spring Batch for efficient daily batch data transfer from Oracle to PostgreSQL databases (part of EDGEAI).
• Improved performance of a critical Python utility script for topology data processing, reducing runtime by 80% (VPNS).
• Enhanced performance of EVS data processing code by refactoring to replace UDFs and explode with flatMapFunction.

Open Source & System Engineering:

• Contributed to an open-source Python SDK enabling developers to utilize the Edge Discovery Service for Verizon's 5G Edge offering (AWS MEC).
• Provided technical leadership and mentorship within the VPNS program, guiding teams on deliverables and managing knowledge transfer during vendor transitions.
• Designed solutions for critical production issues (OOM errors) and led security vulnerability remediation efforts (Fortify, Blackduck) across platforms.

• Significantly improved Oracle database query performance for core applications by implementing strategic table partitioning, enhancing data retrieval efficiency.
• Developed and deployed a critical message processing engine using Jython(Python on JVM) for the VPNS platform, successfully replacing a legacy tool and ensuring reliable handling of network surveillance data.
• Played a key role in large-scale Network Management System consolidation efforts, contributing to application enhancements and gaining expertise in associated networking APIs/protocols (e.g., gRPC, SNMP) relevant to data exchange.
• Enhanced data correlation services for improved network topology traversal and provided technical mentorship to junior team members.

• Developed message parsing and correlation engines for processing network fault alarm messages.
• Contributed to Network Management Systems (NMS) for VoIP/PON networks, including module development for alarm data handling during system consolidation.