MAH

Muhammad Ali Haider

Mechanical Integrity and AI Systems Engineer

Mechanical Integrity · Edge AI Systems · Computer Vision · PINNs

It started with a 2026 MIT video of a drone that flies itself through chaos.

I've been fascinated by this world of intelligent machines ever since.

mah@exubis

Abu Dhabi, UAE

--:--:--

Status

In development

SCALAR by EXUBIS — full-scale asset integrity management software: RBI, IDMS, FFS and PIMs in one platform.

Role

Product Developer

Focus

RBI · IDMS · FFS · Pipelines

Based

Abu Dhabi, UAE

❯ now · scalar · developing the rbi modules

PythonTensorFlowApplied AIRisk Based Inspection (RBI)API 580 / 581API 571Fitness for Service (FFS)Pipeline Integrity (PIMs)Physics-Informed NN (PINNs)Computer VisionSemantic SegmentationCNNsDigital TwinsCorrosion ManagementMechanical IntegrityAPM-MeridiumE2G PMData ScienceEdge AI

Aktuelle Arbeit

Woran ich baue

Venture — ProduktEXUBIS

SCALAR — Asset Integrity Management Platform

End-to-end Asset Integrity Management software built at EXUBIS — Risk-Based Inspection, inspection data management, Fitness-for-Service and pipeline integrity in one platform.

RBI · IDMS · FFS · Pipeline Integrity

Venture — ProduktEXUBIS

STAINS — Statistical Inspection Recommendation

Bayesian network and Monte Carlo simulation over a component's own inspection history, inferring its corrosion rate as a distribution and recommending the next inspection date from it.

Bayesian Networks · Monte Carlo · Corrosion Rates · Inspection Planning

Interessengebiete

Was ich erkunde

On-device inspection models

Running vision models on inspection hardware with no accelerator and a tight power budget, so a technician on a platform gets defect classification in the field rather than back at the office. The open question I care about is how much accuracy survives aggressive quantization.

Edge AI · Computer Vision · Quantization

Vibration-signature fault detection

Comparing learned models against classical signal processing for catching rotating-equipment failure before it happens. Success is predicting a bearing or pump fault far enough ahead to schedule the fix — from the vibration signature alone.

Predictive maintenance · Signal processing · Time series

Physics-informed surrogate models

Where a physics-informed network can stand in for an FEA solver on integrity problems — corrosion growth, remaining-life estimates — and, more importantly, where it quietly cannot. The failure mode I watch for is confident extrapolation outside the training envelope.

PINNs · Simulation · FEA

Model integrity under sensor drift

Deployed corrosion-detection models degrade as cameras and sensors do. I'm interested in flagging that drift from the signal itself — no fresh ground truth — so an integrity call is never made on a silently-decayed model.

Robustness · Monitoring · Calibration

Get in touch

Always glad to talk shop — asset integrity, risk-based inspection, or using AI to make inspection smarter.