Dheeraj Kapur

About Dheeraj Kapur

Own the charter for NVIDIA’s data and observability platforms, taking both from greenfield to company-wide, production-critical infrastructure.ObservabilityBuilt NVIDIA’s unified observability platform from scratch and scaled it to operate at extreme clusters (including multi-tenant tiers), of Vector pipelines, 10s of Loki clusters, and 2.5T+ logs and metrics ingested daily. Established platform safety, isolation, and cost controls and contributed to OSS.Dataverse / Data PlatformDesigned and built the Dataverse platform from first principles: a Trino-based lakehouse with federated and Spark pushdown, OpenMetadata-driven governance, and Hive/Nessie cataloging. Delivered bulk ingestion processing trillions of events/day, CDC pipelines syncing 10s of databases at tens of millions of changes/day, and an Adaptive Router powering 10s of Trino clusters with million’s of routing decisions/day while holding global P95 latency under 30s.Platform & Engineering ImpactShipped new platform components and Dataverse services, plus 13+ OSS and upstream contributions. Established the technical foundation for agentic workflows (MCP, intention routing, anomaly detection, LangGraph) now seeing early internal adoption.

Experience

1. Software Engineer

   NVIDIA

   Jun 2022 — Present · Santa Clarita, CA, US

   Leading Dataverse and Observability platform engineering, with end-to-end ownership of large-scale data and telemetry.Architected and migrated NVIDIA Maglev’s autonomous vehicle data platform, enabling scalable ingestion and analytics for high-throughput vehicle telemetry, sensor data, and downstream analytics systems.Dataverse / Data Lake • Architecting and operating a Trino-based SQL platform over the enterprise data lake • Implemented advanced query pushdown, including Trino→Trino federated pushdown and Trino→Spark delegated execution for optimized cross-engine query processing • Driving metadata, governance, and discovery using OpenMetadata • Managing catalog and table versioning via Hive Metastore and Nessie • Building high-throughput ingestion pipelines using Parquet/Arrow IPC bulk/batch uploads and CDC with DebeziumObservability Platform • Designing and scaling metrics infrastructure using Thanos • Building centralized logging with Loki, powered by Vector ingestion pipelines • Architect-ed multi-region distributed tracing at scale. • Delivering unified visualization, analytics, and operational insights. • Ensuring multi-tenancy, reliability, and performance across all ingestion pipelinesLeadership & Architecture • Owning platform architecture, performance optimization, and long-term technical strategy • Driving software engineering best practices across distributed systems and data platforms • Mentoring engineers and leading cross-team initiatives spanning data, observability, and infrastructure

Education

• Guru Gobind Singh Indraprastha University

  B. Tech, Electronics and Communication Engg

• Delhi College of Engineering

  M. Tech, Electronics and Communication Engg

Skills

• Cluster
• Python
• Storm
• Automation
• Enterprise Software
• Data Structures
• Open Source
• Git
• Databases
• Apache
• Cloud Computing
• Shell Scripting
• Php
• Web Applications
• Java
• Linux Kernel
• Linux
• Hive
• Hadoop
• Perl
• Hbase
• Mysql
• Distributed Systems
• Mapreduce
• System Administration
• Bash
• Unix
• Rest
• Mobile Devices
• Scalability
• High Availability