Hello, I am Dickson πŸš€

Full Stack Computer Vision Engineer

I make models small, fast, and efficient.

See my projects ↑
Something personal

About me


I am a seasoned deep learning data scientist with a passion for using cutting-edge technology to drive real-world results. With years of experience in designing, building, and deploying advanced deep learning models, I’ve also helped companies design end-to-end deep learning solutions and deploy them into production.

Prior to joining the industry, I was an academic with a proven track record. I’ve published in reputable journals including Nature, Elsevier, and IEEE Transactions. See my publication list here.

I’m an introvert - INFJ according to the MBTI personality test. I enjoy reading self-development books on psychology, nutrition, biohacking and productivity. Among my favorites are Deep Work, Deep Nutrition, Quiet, Ultralearning, The Cancer Code, The Bulletproof Plan to Age Backward, and The 7 Habits of Highly Effective People. See all the books I’ve read here.

Other than reading I enjoy tinkering with electronics and coding in my free time. I’ve made Raspberry Pi-based electronic locks, Telegram bots to monitor crowd attendance, an Android app to broadcast live stream videos, and fixed old electronic board problems on televisions, ceiling fans, etc.

I enjoy creating contents on practical deep learning projects - some of which had garnered over 400,000 views in the course of 6 weeks.

Support me in pushing out more high-quality ML contents.

Buy Me A Coffee

figure-svg
about-img

I have a unique blend of academic and industry experience.

What I bring to the table –

🟩 10+ years of modeling DL-based time series and computer vision algorithms.

🟩 4+ years of helping companies deploy DL models into production.

🟩 10+ years of academic journal and conference publication track record. Full publication list here.

🟩 Named world’s Top 2% Scientist by Stanford University in 2023.

🟩 Proven ability to think from first principles in designing high quality, scalable DL-based solutions.

🟩 Unique combination of technical know-how and communication to a non-technical audience through written and spoken mediums.

🟩 Influential social media presence. My recent Twitter and LinkedIn posts generated over 400k impressions in 6 weeks without spending a dime on ads. Check out my best Tweets and LinkedIn posts.

🟩 Multilingual – English, Mandarin, Malay, basic Korean.

🟩 Trained in Neuro-Linguistic Programming (NLP) and Emotional Freedom Technique (EFT) Tapping.

background-pattern
How can I help

Services

Data & Model Development, Writing, and Consultation

  Chat on WhatsApp   Chat on Telegram
ui-ux

Consultation

1-hour personal video call session with me to discuss anything related to your project.

ui-ux

Writing

I will write one technical blog post on your tech.

ui-ux

Data Engineering

I will clean your image dataset, label them and craft a balanced train, validation and test set.

ui-ux

Model Development

I will train a computer vision model for you with a provided dataset.

ui-ux

Optimization

I will optimize your model for low-latency inferencing (OpenVINO, DeepSparse, ONNX, TFLite, Torchscript).

ui-ux

Front End Application

I will develop a Flutter app (Android/iOS/Web) for your computer vision model.

A summary of

My Expertise

Tech stack –

🟩 Python, C, Assembly, Dart.

🟩 Docker, Jupyter, VSCode, Git.

🟩 DL framework – Tensorflow, PyTorch, Keras, Fastai.

🟩 DL experiment – Weights & Biases, Tensorboard, CometML, Optuna, Neural Network Intelligence.

🟩 DL deployment – OpenVINO, TensorRT, ONNX, TFLite, TorchScript, Hugging Face Hub, DeepSparse.

🟩 Database – PostgreSQL, MySQL, Firebase, Hugging Face Datasets.

🟩 UI – Flutter, Gradio, Flet, Streamlit.

🟩 Edge Device – Android/iOS Device, Raspberry Pi, NVIDIA Jetson, Arduino/ESP32/PIC/AVR MCU, Coral Edge TPU, Intel Neural Compute Stick.

🟩 Current GitHub Stats –




2018-2022

Ph.D. in Engineering

My Ph.D. work explores the effectiveness of using deep learning models to estimate state-of-charge (SOC) in the batteries of hybrid electric vehicles. The study conducts various in-depth comparative analyses of state-of-the-art deep learning methods applied to SOC estimation. The goal of the study is to develop a novel estimation algorithm that can accurately estimate the remaining charge from the drivers’ driving patterns. This work proposes the use of a self-supervised Transformer model and yields low error metrics providing a promising alternative to conventional SOC estimation models. See our published Nature paper.

2012-2015

Master’s Degree in Electrical Engineering

My Master’s Degree work explores the idea of recognizing the behavior of humanoid robots using a Long Short-Term Memory (LSTM), a variation of recurrent neural networks. The LSTM network is shown capable of classifying robotic maneuvers from joint angle data. See our published paper and demo video.

2007-2012

Bachelor of Electrical and Electronics Engineering (Honours)

I graduated first class with a cumulative GPA of 3.71/4.00. My bachelor’s thesis explores the idea of using pulse width modulation (PWM) techniques in combination with PID control algorithms to control the torque of a brushed DC motor. I fabricated a custom printed circuit board (PCB) to proof the working principle of the motor driver. See our published paper.

March 2023 - Present

Visual Layer Inc.

Machine Learning Engineer.

Helped maintain the open access visual data curation software - fastdup. Wrote blogs as a representative of Visual Layer. Assisted in community building and question handling on Slack/Discord.

July 2022 - December 2022

ZenML GmbH

Developer Advocate.

Developed social media content strategy for ZenML’s Twitter and LinkedIn. Wrote blogs and gave podcasts/talks as a representative of ZenML. Assisted in community building and question handling on Slack.

May 2016-November 2022

National Energy University, Malaysia.

Lecturer.

Taught undergraduate and diploma level courses. Primary subjects taught include Microprocessor Systems, Digital Logic Design and Random Process. I was also featured once on Malaysia’s national TV showcasing my efforts on robotics education to the younger generation.

2012-2016

Center for Advanced Mechatronics and Robotics.

Research Engineer.

Responsible to design boiler header inspection robot, algorithms for P3AT line following robot, algorithms for behavior classification with LSTM on NAO humanoid robot, develop learning modules for workshops on Arduino microcontroller.

background-shape
figure-svg
skill-img
What I do best

Why hire me for your next project?

Throughout the years in academia with tight industry collaborations, I have developed a unique blend of skills and expertise in both software and hardware.

Deep Learning (Tensorflow/PyTorch)
90%
Model Optimization (OpenVINO/TensorRT/ONNX)
85%
Edge Computing (Nvidia Jetson/Intel VPU)
80%
Microcontrollers (Arduino/Raspberry Pi)
75%
Frontend (Flutter/Streamlit/Gradio)
70%
Databases (PostgreSQL/MySQL)
60%
Cloud Services (AWS/Azure/Google Cloud)
50%
blog shape