Skip to portfolio

Cloud, infrastructure,
and what I’m working on.

A tool that sizes GPU setups for us.

Turning a customer’s workload into a GPU design was slow. I built a configurator to do the first pass.

The need

Every AI proposal needed GPU sizing and a one-page summary for the customer. Doing that by hand ate days of the presales team’s time.

My part

I built it with a self-hosted LLM endpoint, Python and Hermes agent skills. You give it the workload and it works out the GPU resources, sizes the workload and produces the infographic.

How I went about it

I ran the model on a self-hosted endpoint rather than an outside service, and kept the sizing and the customer-facing material in one tool.

The outcome

Presales designs that used to take more than four days now take about two hours. I also showed it at the hackathon session of the Dell × NVIDIA AI Advisors Workshop.

Internal result. Turnaround depends on how big the solution is.

CLOUD · GPU · IT INFRASTRUCTURE

Zulficar
Mohd Yaakub.

Solution Engineer

AI Cloud & Infrastructure

I build the infrastructure AI runs on: private clouds, GPU platforms and the self-hosted tools on top. I’m usually the one in the client meeting who then goes and builds the thing.

Based in Selangor, Malaysia7+ years in IT infrastructure

SELECTED WORK / 01—06

A few things
I’ve worked on.

AI solution sizing, sovereign cloud design,
and the infrastructure behind them.

EXPLORE MY WORKFrom AI tools to private clouds

Portrait of Zulficar Mohd Yaakub

Selangor00:00 MYT

WHAT I’M WORKING ON

Right now.

GB10 + vLLMRunning models for our internal tools
GPU configuratorSizing AI setups for presales
RobustGXNode services and CI/CD

EXPERIENCE AT A GLANCE

7+

years in cloud and
IT infrastructure

1,184 linesRequirements · sovereign AI cloud 30+ nodesPrivate cloud · 100+ TB 2,000+ usersMulti-campus phone system

BIGGEST WIN LATELY

4+ days → 2 hours

How long a presales GPU design took, before and after the configurator I built.

Internal result · depends on solution sizeSee how it works →

HOW I WORK

Design it for whoever gets the 2am call.

I start by finding out what’s already running and what’s off the table.

CERTIFICATIONS & TRAINING

What I’ve been learning.

NVIDIA AI Advisor
Rafay GPU Cloud
MinIO Sales
ISC2 CC

See all credentials

WHAT I WORK WITH

The tools behind the projects.

AI & inference →

NVIDIA platforms · vLLM · GPU sizing

Cloud & virtualisation →

OpenStack · Kubernetes · Proxmox · VMware

Automation →

Python · Ansible · Terraform · GitLab CI/CD

OUTSIDE WORK

Chair's Lab

My personal blog, for whatever I find interesting and want to share, plus the open source projects I want to put out there.

BlogOpen source

Work in progressVisit chairlab.xyz ↗

WHERE I’VE WORKED

From support to AI infrastructure.

2023 — PRESENT

Solution Engineer

Robust HPC · AI cloud & infrastructure

2021 — 2022

Head of Unit — IT Infrastructure

Confidential employer · Private cloud & delivery

2020 — 2021

Data Security Analyst

Confidential employer · Logging & data infrastructure

2019 — 2020

Support Technical Engineer

Astiostech · Communications & IT operations

Request my résumé

GET IN TOUCH

Let’s talk
infrastructure.

Working on an AI platform, GPU cloud or infrastructure project? Tell me what you’re planning.

Request résumé LinkedIn ↗