Mechatronics Engineer · Applied AI & ML · Engineering to Business Impact

Where engineering meets the decisions behind it.

I build and evaluate machine learning systems for industrial hardware, and I care just as much about why they get built: the cost case, the rollout, and whether anyone actually trusts the output.

That double interest is the thread through my work so far — deep technical projects on one side, go-to-market and public-sector work on the other.

B.Eng. Mechatronics, THWS Germany · Available in Malaysia from January 2027

Afiq Zulfan

0

Audio samples collected and processed for model training

0

Machine learning detection approaches designed, compared and benchmarked

0

Competitors benchmarked for product positioning

€0

Budget for an approved robotics prototype

About

What I bring to the table

I recently completed a B.Eng. in Mechatronics at THWS in Germany, with most of my practical experience gained inside one of Germany's largest technology groups, the kind of environment where a model only matters once it runs on real hardware and someone can sell it.

01

Technical depth

Python, TensorFlow and signal processing applied to real sensor data, down to quantised INT8 inference on a microcontroller.

02

Structured problem-solving

I break an open-ended problem into testable options, compare them honestly, and recommend one I can defend, including when the honest answer is that it does not work yet.

03

Technical to non-technical

Competitor analysis, sales enablement material and technical guides written for colleagues who need clarity, not jargon.

04

Three languages, two working cultures

Malay, English and German. Four years studying and working in Germany, now bringing that back to Malaysia.

Focus areas

What I'm building toward

Four areas I keep returning to, in projects and in reading.

01

Applied AI & machine learning

Classification and anomaly detection on sensor data — model selection, evaluation, and deployment on constrained hardware.

02

Industrial IoT & predictive maintenance

Sensor-to-cloud connectivity, condition monitoring, and turning vibration or acoustic trends into maintenance decisions.

03

Technology strategy & consulting

The side I want to grow into: translating what the technology can do into a business case a client can act on.

04

Public policy & the public sector

I follow AI and digitalisation policy closely, and spent a fellowship inside Malaysia's foreign ministry. I'd like to stay close to how the public sector adopts technology.

How acoustic leak detection works

Switch the pipe condition, then watch the signal and the classifier.

No leak detected confidence 4%

Illustrative simulation of the detection concept, not live model output. A leak adds continuous high-frequency energy on top of normal machine noise. That difference is what the model learns.

Education

Academic foundation

Oct 2022 — Oct 2026

B.Eng. Mechatronics

Technical University of Applied Sciences Würzburg-Schweinfurt (THWS), Germany

  • Specialisations: Human–Machine Interaction (HMI), Automated Systems, Robotics and Production
  • Thesis title: SITRANS MS200 microphone data processing & Edge AI leak detection — graded 1.0 (Best)

Mar 2022 — Aug 2022

Intensive German language course — CEFR B1

Institut für Internationale Kommunikation e.V. (IIK Berlin), Germany

Jul 2020 — Feb 2022

German A-Level Preparatory Programme (GAPP)

German-Malaysian Institute (GMI), Selangor, Malaysia

  • Subjects: Mathematics, Physics, Chemistry and German language
  • Pre-study engineering training (Vorpraktikum), Sep 2021 – Feb 2022: benchwork, milling, turning, grinding and welding; electrical installations, electronics and digital systems; programming in C, pneumatics, hydraulics and technical communication

Feb 2018 — Nov 2019

Sijil Pelajaran Malaysia (SPM)

MRSM Pengkalan Hulu, Perak, Malaysia

  • Sports & Recreation Bureau, Student Representative Council
  • Volleyball and Scouts Jamboree
  • International Innovation, Design and Articulation (i-IDeA) competition

Work

Experience

Three roles at Siemens Digital Industries, plus a fellowship at Malaysia's Ministry of Foreign Affairs.

Apr 2026 — Sep 2026

Bachelor's Thesis Student

Siemens AG — Karlsruhe, Germany

Project: Edge AI for Industrial Acoustic Leak Detection

Situation

Compressed-air leaks waste energy continuously, but are usually found only during manual walk-around checks.

Approach

Trained on over 15,000 audio samples and compared three detection approaches (energy threshold, MLP, DS-CNN), then quantised the winner to INT8 and built new firmware that runs the model on the microcontroller inside the multisensor itself — acquisition, spectral preprocessing and inference, all on the device.

Impact

The learned model showed a 23.9% relative improvement in F1 score over a conventional energy-threshold baseline, and the complete pipeline ran reliably on the device through repeated live acquisition cycles.

PythonTensorFlowTFLite MicroEmbedded firmwareSignal processingINT8 quantisation

Feb 2026 — Mar 2026

Working Student, Go-to-Market Support

Siemens AG — Karlsruhe, Germany

Siemens Smart Condition Monitoring (SCM IQ)

Situation

The SCM IQ team needed sharper positioning material in a crowded condition-monitoring market.

Approach

Benchmarked 10 competitors against the SCM IQ portfolio and turned the findings into sales enablement material and technical product positioning.

Impact

Adopted by sales colleagues and business developers as reference material in their client conversations.

Competitor analysisProduct positioningSales enablementMarket research

Sep 2025 — Jan 2026

Intern, Predictive Maintenance & Smart Condition Monitoring

Siemens AG — Karlsruhe, Germany

Situation

Siemens' condition monitoring sensors operated solely within the SCM IQ environment. The team was opening this up, giving customers the freedom to route their sensor data to the platform of their choice.

Approach

Took the user's perspective on each new release, validating MS200-to-CC220 gateway connectivity over MQTT across four integration environments, and relayed every finding back to the development team for the next iteration.

Impact

Raised more than 10 configuration issues that were fixed before release, then documented the step-by-step guide, which became the foundation of the official documentation now used by regional colleagues.

MQTTSenseyeCondition monitoringIntegration testingTechnical writing

Aug 2023 — Oct 2023

Perdana Fellow, Office of the Minister of Foreign Affairs

Wisma Putra — Ministry of Foreign Affairs, Malaysia

Public Communication & Public Diplomacy Division

Situation

The ministry's Merakyatkan Diplomasi agenda aimed to bring foreign policy closer to ordinary Malaysians, with no youth channel in place.

Approach

Worked under the Public Communication and Public Diplomacy Division, supporting the agenda and coordinating engagements with the Minister's Office, and co-founded the Wisma Youth Attaché volunteer programme.

Impact

An inaugural cohort of roughly 30 volunteers, officially launched by the Minister and senior leadership at the first-ever Hari Wisma Putra, giving people-centred diplomacy a standing channel to the public.

Public diplomacyPolicy researchProgramme designStakeholder management

Projects

Nov 2024 — Jan 2025

Robot for maintaining a green façade

THWS × Center for Applied Energy Research

Situation

Green façades improve building efficiency but are awkward and costly to maintain by hand at height.

Approach

Led a team of five through a proof of concept using the VDI 2221 systematic design methodology for concept development and evaluation, then built Python motion control on a Raspberry Pi to automate the robot's movement.

Impact

Delivered a working prototype on a budget under €160, documented the full design process and presented the findings. The concept was approved on handover and carried into the next development phase.

PythonRaspberry PiVDI 2221Motion controlTeam leadBudget management

Dec 2024

AI & IoT-based smart traffic management

THWS × Green City Schweinfurt

Situation

Schweinfurt's green-city initiative was looking at how sensing could ease urban congestion.

Approach

Built an IoT traffic control prototype on Raspberry Pi with Python, MQTT and Node-RED, and integrated an image classifier to recognise defined traffic scenarios.

Impact

Demonstrated automated signal switching with congestion alerts visualised live on a dashboard, end to end from camera to notification.

Node-REDMQTTImage classificationRaspberry PiDashboarding
A model is only finished when someone can act on it.
— how I think about engineering work

Say hello

Let's talk

Open to graduate roles in engineering, data and technology consulting. Based in Kuala Lumpur from January 2027.

Location

Kuala Lumpur, Malaysia