Engineering. With perspective. People. Platforms. Possibility.

Engineering leader · AI infrastructure

Hi, I’m Sreerag Sreenivasan.

Build the systems.
Grow the teams.
Make an impact.

I lead teams delivering GPU infrastructure software, from platform strategy to customer acceptance. Connecting technical depth, clear direction, and people who care about the work.

Explore my work Request résumé
Seattle area, Washington / Building at Oracle
Portrait of Sreerag Sreenivasan against a soft neutral studio background
People. Platforms. Possibility.Engineering with a bigger picture.
Grounded in engineering. Growing impact through teams. Scroll to explore
A career built across Oracle teradata. TATA CONSULTANCY SERVICES 12+ years of building

01 Selected impact

The work.
The difference it made.

Three changes. Delivered by teams. Enabled through leadership.

AI infrastructure

Higher GPU availability.
Faster recovery.

FromManual repairs
ToTargeted automation
My role
  • Strategy
  • Team formation
  • Partnerships
Read the story
The challenge
Manual repair work and recurring failures limited usable GPU capacity and pulled engineers into escalations.
How I enabled the team
I set fleet-health priorities, formed a focused remediation team, and aligned engineering and operations on ownership and handoff. Teams turned recurring repairs into reusable automation.
What changed
Teams improved GPU availability, reduced recovery time and manual intervention, and established repair capabilities that operations and adjacent teams could maintain and extend.

Cloud identity

More reliable access.
Fewer disruptions.

FromReactive fixes
ToBuilt-in resilience
My role
  • Release strategy
  • Team alignment
  • SDK adoption
Read the story
The challenge
A critical identity service supported many dependent services. Improving reliability required changes to releases and to how clients handled failures.
How I enabled the teams
I set release-management priorities and led a cross-team adoption effort for client safeguards, aligning service owners around circuit breakers and bounded retries.
What changed
Teams reduced major incidents and shortened mitigation time, strengthening reliability across dependent services. Work on region-build dependencies also brought identity services online sooner.

Analytics platforms

Faster access.
Lower cloud cost.

FromCumbersome provisioning
ToSelf-service workspaces
My role
  • Roadmap
  • Delivery
  • Customer feedback
Read the story
The challenge
Data scientists needed on-demand workspaces and access to their preferred tools without a cumbersome provisioning process.
How I enabled the team
I led delivery of stateless, containerized services for a multitenant analytics platform and used customer feedback to guide the roadmap. Earlier, I built REST services and JupyterLab integrations for self-service workspaces.
What changed
Workspace provisioning became self-service and faster, while the shared-service approach reduced cloud costs. Customer feedback continued to guide the platform roadmap.

02 How I lead

Build leadership
that lasts.

I lead through engineering managers and connect roadmaps, resource priorities, and operating standards across teams and sites.

Leadership that extends beyond one team

Engineers grew into managers.

Engineers I sponsored moved into management within my team and adjacent organizations.

  1. Recognize potentialJudgment & initiative
  2. Delegate real scopeResponsibility & support
  3. Build independenceLeaders developing teams
My role in the transition
  • Coach delegation, feedback, and performance conversations.
  • Connect growth opportunities with real engineering needs.
  • Support new managers as they establish team ownership.

An operating model teams can sustain

Recurring issues became owned work.

Weekly reviews connected operational problems to named engineering improvements.

  1. Make work visibleSignals & customer impact
  2. Name the ownerPriorities & accountability
  3. Make it repeatableReviews & operational handoff
My role in the operating model

I established priorities, escalation paths, and a review cadence. Engineers owned the improvements and connected technical decisions to customer impact; useful automation became part of sustainable operations.

Engineering, operations & customer teams

Shared checks. Clear handoffs.

I negotiated validation on both sides of repair handoff; teams later partially automated the process.

  1. Agree on the outcomeCustomer needs & quality
  2. Resolve the handoffsShared checks & clear owners
  3. Improve togetherValidate, learn & automate
My role across teams

I aligned delivery priorities, surfaced dependencies, and resolved ownership gaps. Customer feedback informed engineering priorities, while resource planning supported delivery needs.

03 Engineering judgment

Technical depth.
Shared direction.

Architecture choices teams can deliver and operate.

The prototype: remote remediation
Java / SpringRemediation service
PythonGPU remediation scripts

Remote invocation of GPU remediation actions.

Discuss the design

Repair automation

From direction
to shared capability.

  1. I set directionFocused approach · prototype · team formation
  2. The team deliveredProduction remediation framework
  3. Operations took ownershipHandoff · extension · adoption by other teams
  • Less manual repair
  • Faster recovery
  • Broader adoption
The tradeoff behind the approach

Recurring remediation was consuming engineering time. I chose a focused path to reach operations sooner and made operational ownership part of the plan.

Alternative

Broader platform integration

A longer path to the immediate repair capability.

The choice

Automation in the repair platform

Earlier delivery, with a framework operators could extend.

My contribution
Technical direction, team formation, and operational handoff, supported by an initial Java/Spring prototype remotely invoking Python remediation scripts. The team carried the framework into production.
Another decision: make failure patterns usable
I established structured error signatures as a shared input and aligned customer teams on providing them. Combined with AI-assisted analysis, this helped engineers find recurring patterns and shorten investigation.

Explore the decision

When is GPU capacity
ready for customers?

Explore how technical evidence and ownership
shape a release decision.

GPU readiness lab Illustrative model · No live data
Try a scenario
Change the evidence
Release decisionHeld
Illustrative capacity path
GPU nodesCheck health FabricQualify fabric Workload eligibilityHeld
Refresh evidenceHandoff authorized

1 of 4 conditions met

Hold for
qualification.

A completed repair does not prove GPU and host health. Re-run validation before relying on this capacity.

Still required: node validation, fabric qualification, current evidence.

Next collaborationPlatform engineering + Qualification
Hardware repaired → evidence renewed → capacity qualifiedEvery required condition matters.

An engineering foundation

  • REST services
  • JupyterLab
  • Database scale-out
Explore earlier code

04 Where I’ve built

Built on
experience.

Request full résumé

Foundations

M.S., Computer ScienceSan Diego State University · 2018

B.Tech., Electronics & CommunicationUniversity of Calicut · 2014

Oracle AI Infrastructure & Cloud Identity

Engineering leadership across GPU infrastructure and cloud identity, spanning platform direction, qualification, deployment, repair, and customer acceptance.

AI infrastructureCloud identityOrg leadership
Teradata Cloud Analytics & Developer Platforms

Hands-on software development and team leadership across self-service analytics, developer APIs, cloud workspaces, database services, and high availability.

AWSDeveloper platformsTeam leadership
Tata Consultancy Services Enterprise Software

Built workflow automation, REST integrations, and client-side tooling for enterprise applications, alongside improvements to testing and continuous integration.

Workflow automationREST APIsCI/CD

05 Engineering roots

Curiosity, in code.

Earlier experiments. Public source code. A builder’s foundation.

Explore GitHub

4 projects

Web application

Campus book exchange

A student marketplace for used textbooks.

Source

Android · Firebase

Geekspad

Tech meetups and conversations for developers.

Source

Data science · Machine learning

What makes a movie work?

Movie data explored through machine learning.

Source

Arduino · Embedded systems

A hand, built from scratch

Finger flexion translated into robotic movement.

Source

06 Start a conversation

Good things start
with hello.

Call me Sree. Let’s talk about developing engineering leaders, setting technical direction, and helping teams deliver reliable infrastructure.

Let’s connect
Connect professionallyLinkedIn Explore the codeGitHub The full story, by emailRequest résumé

View source on GitHub