01
GPU Infrastructure
Observability and fault detection for distributed GPU environments powering AI workloads.
GPU ComputeTelemetryReliability
Building at OCI · Seattle
Shubham
Gulati
Principal Engineer building GPU compute, AI infrastructure & distributed systems.
I design and build production systems for AI workloads — with a focus on GPU observability, distributed data processing, reliability, and high-scale cloud infrastructure.
Shubham Gulatiprincipal engineer · seattle, wa
current_focus
GPU observability
fault detection
GPU observability
fault detection
stream
Object Storage → Flink
→ Kafka
Object Storage → Flink
→ Kafka
01 / What I build
Infrastructure that stays boring when the scale gets interesting.
My work sits at the intersection of distributed systems, GPU infrastructure, AI platforms, and reliability engineering.
02
Distributed Systems
High-scale systems designed around availability, capacity, concurrency, and operational simplicity.
FlinkKafkaDynamoDB
03
AI Infrastructure
Developer-facing AI platforms, agentic tooling, MCP integrations, and NL-to-SQL.
MCPAgentsNL-to-SQL
$ current-focus
→ GPU compute observability
→ streaming fault detection
→ production reliability
$
→ GPU compute observability
→ streaming fault detection
→ production reliability
$
02 / Experience
Production systems, end to end.
Oct 2025 — Present
Principal Member of Technical Staff
Oracle Cloud Infrastructure
Building GPU compute observability for OCI AI infrastructure, processing telemetry through Apache Flink to detect actionable infrastructure faults and publishing events to Apache Kafka.
Aug 2022 — Oct 2025
Senior Member of Technical Staff
Oracle Cloud Infrastructure
Tech lead for Secure Big Data Gateway, including architecture, production launch, capacity planning, canarying, observability, and cross-team reliability work.
Feb 2020 — May 2022
Software Engineer — DynamoDB / JournalDB
Amazon Web Services
Built distributed database infrastructure including multi-tenancy, automation, and cross-account VPC connectivity.
May 2016 — Aug 2018
Software Development Engineer
Zomato
Built microservices and internal platforms for merchant and content systems.
03 / Selected work
A few systems worth talking about.
GPU INFRASTRUCTURE · STREAM PROCESSING
GPU Fault Detection Pipeline
Streaming observability pipeline for GPU clusters.
Object Storage / Telemetry
Apache Flink / Stateful Processing
Fault Detection / Signal Extraction
Apache Kafka / Customer Workflows
DURABLE EXECUTION
Agent Execution Runtime
Fault-tolerant execution with retries, event-sourced durability, deterministic replay, and OpenTelemetry tracing.
LLM INFRASTRUCTURE
KV Cache Manager Simulator
PagedAttention-style allocator with copy-on-write, prefix sharing, eviction policies, and benchmarking.
04 / Stack
Tools change. Systems thinking compounds.
Languages
Java
Python
Go
Rust
SQL
Python
Go
Rust
SQL
Distributed
Apache Flink
Apache Kafka
DynamoDB
Hadoop / HDFS
ZooKeeper
Apache Kafka
DynamoDB
Hadoop / HDFS
ZooKeeper
Cloud & Infra
OCI
AWS
Kubernetes / OKE
Kubelet
Docker · Terraform
AWS
Kubernetes / OKE
Kubelet
Docker · Terraform
AI & Reliability
GPU Compute
GPU Observability
MCP
OpenTelemetry
SLOs
GPU Observability
MCP
OpenTelemetry
SLOs
05 / Education
Computer science, with a systems bias.
University at Buffalo
M.S. in Computer Science · 2018–2019
Punjab Engineering College
B.E. in Information Technology · 2011–2015
06 / Contact
Let’s build something difficult.
Interested in distributed systems, AI infrastructure, GPU platforms, or reliability engineering? Reach out.