Kuala Lumpur, Malaysia
Amir Ashraf Izham — Lead Data Scientist
Engineering the AI future
I build AI that turns complex data into strategic decisions.
Building AI for palm oil | Mechanical Engineer at heart
About
From engines to intelligence
0+
years in industry
engineering to AI
0
industries shipped in
energy, agriculture, public sector
0
specialists led
across data science, AI engineering & MLOps
I started in mechanical engineering—BEng at Nottingham, MSc at Manchester—before trading physical systems for digital ones. That pivot sparked years of deliberate unlearning and relearning.
Along the way, I learned to ship products at PETRONAS Digital, architect cloud ecosystems at Microsoft, and scale enterprise AI/ML systems at Accenture and SD Guthrie.
Each move meant trading an old instinct for a better one: swapping “perfect before shipping” for rapid iteration, and “build it all myself” for building through a team.
The mission behind it all? Bringing data, cloud, and AI together to solve real-world problems.
Education
Built on mechanical engineering
University of Nottingham
Malaysia
BEng Mechanical Engineering
2:1
- Undergraduate coursework in Robotics, Dynamics and Control Systems.
- Dissertation: Dynamics and Vibration of a Hand-Powered Drill.
University of Manchester
UK
MSc Mechanical Engineering Design
Distinction
- Postgraduate coursework in Vibration, Acoustics and Mechanical Design.
- Dissertation: Thermo-Mechanical Analysis of a Silicone Shoe Insole.
Career
Four companies, one mission: turning data into decisions
Apr 2025 – present
SD Guthrie
Lead Data Scientist
- Built probabilistic LightGBM yield forecasts for 121 estates, 18 months ahead, at 5% MAPE — scaled to 2 more countries.
- Forecast the mill's Oil Extraction Rate with XGBoost at 1.5% MAPE.
- Led an AI data analyst agent on Databricks Genie: time-to-insight down 93% (~30 → ~2 min).
- Scaled the team from 2 to 5 and moved it to Agile + CRISP-DM.
Aug 2023 – Apr 2025
Accenture
Data Science Consultant
- Optimised the coke yield of a delayed coker unit: +52% R², +2% yield.
- Built an MLOps CI/CD/CT/CM architecture that cut training time by 66% (90 → 30 min).
Oct 2022 – Jun 2023
Microsoft
Associate Cloud Solution Architect
- Led data cleansing of Azure services for the Malaysia West data centre repatriation.
- Designed N-tier, microservice and hybrid architectures with the Azure Well-Architected Framework.
- Awarded the Growth Mindset Award.
Jun 2018 – Jun 2022
PETRONAS Digital
Software Engineer
- Led the architecture of a real-time ML recommendation app for a gas plant on AWS; scaled to 300% more users with no performance loss.
- Automated a refinery simulation engine predicting crude properties 75 days ahead — from 4,320 manual hours to 10 minutes.
- Led the delivery of a heat exchanger monitoring app, unlocking RM 0.4M/year in cost avoidance.
- Pioneered DevSecOps with CI/CD, SonarCloud and Docker on Azure DevOps.
- Awards: APIC Gold 5 Stars, RISTeX Gold Medal & Industry 4.0, IChemE Global Award, GasTech Award.
Featured work
Built for the real world
Personal projects
Built after hours, out of pure curiosity
Skills
Equal parts cloud, data and AI
The daily toolkit. Boring tools, used well, beat shiny tools used badly.
Credentials
- Azure Solutions Architect Expert (AZ-305)
- Azure Data Scientist Associate (DP-100)
Contact