> whoami

Piyush Dubey

I build distributed backend systems, scalable platforms, and practical AI-powered tools with a focus on reliability, performance, and developer productivity.

./about
./selected_impact
./notes
connect_linkedin
git_clone_repo
</>
AI
API
0101_
01.

About

> Loading professional summary...

I am a Bangalore-based principal engineer with 13+ years of experience across Oracle Cloud, Microsoft, Adobe, and earlier enterprise platforms. My work sits at the intersection of distributed backend systems, infrastructure, and AI integration, with a bias toward systems that need to scale cleanly and remain operable under pressure.

At Oracle OCI, I build quota and delivery infrastructure: a Redis-backed distributed rate limiter scaled to 10K QPS, a core microservice re-architected from single-region to multi-region for 99.9% uptime, and Kafka-driven CI/CD event services with cross-region durability backed by Oracle Database. I also built MCP-based AI agent tooling and CLI workflows that are now used across multiple engineering teams.

Before Oracle, I designed architecture for Microsoft 365 Groups AI Platform powering Copilot search and Q&A for 220M+ monthly active users, improved backend latency by 70%, and shipped enterprise lifecycle features with measurable cost savings. At Adobe, I built translation and eventing systems for 31 locales at 50K QPS. Along the way, I have mentored 40+ engineers, led architecture reviews, and kept a strong focus on turning complex systems into something teams can actually move with._

./capabilities --list

  • Java
  • Go
  • Python
  • C#
  • Spring Boot
  • Micronaut
  • Dropwizard
  • FastAPI
  • Kafka
  • Redis
  • PostgreSQL
  • Oracle DB
  • Cassandra
  • MongoDB
  • Kubernetes
  • Docker
  • Terraform
  • LangChain Agents
  • MCP
  • Semantic Search
  • Vector Embeddings
  • OpenAI Codex SDK
  • Claude
  • OCI
  • Azure
  • AWS
  • Microservices
  • Event-Driven Systems
  • Multi-Region HA
  • Rate Limiting
02.

Selected_Impact

A snapshot of the systems I have led across multi-region infrastructure, AI platforms, and backend services that operate at production scale.

Distributed Platform Engineering at Oracle

Oracle Cloud Infrastructure | Principal Member of Technical Staff

  • Built a distributed rate limiter for user and AI agent quota management, scaling the service to 10K QPS with Redis-backed shared state.
  • Re-architected a core Oracle Cloud microservice from single-region to multi-region, improving availability to 99.9% for critical infrastructure paths.
  • Built Kafka-based event streaming for CI/CD with cross-region support and durability guarantees backed by Oracle Database integration.

Agentic AI Tooling for Engineering Teams

Oracle Cloud Infrastructure | AI Tooling + Technical Leadership

  • Created MCP server and CLI-based AI agent harnesses used across multiple engineering teams for agentic AI development and support workflows.
  • Focused on practical adoption by building reusable tooling around LangChain agents, OpenAI Codex SDK, Claude, and internal developer workflows.
  • Drove architecture review committees, AI forums, and mentorship programs spanning 40+ engineers across the organization.

Microsoft 365 Groups AI Platform

Microsoft | Senior Software Engineer

  • Designed architecture for the Groups AI Platform powering Copilot search and Q&A for 220M+ monthly active users.
  • Improved API retrieval and collaboration latency by 70% through targeted backend and platform optimizations.
  • Led cross-team execution across Outlook, Teams, and SharePoint while shipping lifecycle capabilities that delivered about 10% subscription cost savings for Tier-1 enterprise customers.

Adobe, Globalization, and Early Foundations

Adobe + TCS | Computer Scientist / Software Engineer

  • Built real-time translation microservices supporting 31 locales at 50K QPS, helping accelerate international product launches.
  • Designed Kafka-based event pipelines with retries, batching, and DLQ handling, plus globalization services that reduced localization cost and time-to-market.
  • Built LINE messaging integrations that reached 60% adoption in the Japan market, and earlier developed healthcare data pipelines that improved reliability for patient-doctor alignment systems.
03.

Notes

I write about distributed architecture, backend systems, AI agents, and the engineering trade-offs behind building reliable platforms. Most of it comes from real production work: multi-region services, event-driven systems, developer tooling, and leading architecture across teams.

read_on_linkedin