So, what made Edward Edward

I started my UK journey at UCL Electronic and Electrical Engineering. The program gave me a strong foundation in engineering, science, and programming. Particularly, the interdisciplinary engineering challenges known as "Scenarios" exposed me to a diverse range of expertise and enhanced my communication skills. I discovered my passion for programming after exploring various academic fields. In the final year, I chose a machine learning module to solidify my foundations and an accounting module out of personal interest. My final dissertation focused on emotional voice conversion. which involved audio analysis, CNN, and CycleGAN.

With a deep interest in machine learning, I pursued a MSc degree in UCL Machine Learning. During my studies, I gained invaluable hands-on experience with various aspects of machine learning, such as NLP, information retrieval, data mining, machine vision, and supervised/unsupervised learning, to name just a few. For my final dissertation, I executed semantic segmentation on aerial imagery, where I improved the performance by enhancing boundary information. This project involved computer vision, U-Net, deep learning, and CNN, and was a resounding success.

Ed's experiences

  • Intern @ CYENS Centre of Excellence
    • Project on semantic segmentation. Based on U-Net and aerial images.
    • Improved the model performance by innovative training and post-processing techniques.
    • An end-to-end project including data gathering, data cleaning, fine-tuning, analyzing, post-processing.

  • University Representative Assistant @ Intake University Fair
    • Assisted in the communication between participating students and university representatives. Provided translation for the students if necessary.

  • Course Academic Representative @ UCL
    • Collected, organised, and presented the feedback of over 100 students.
    • Consolidate students' views and discuss with teachers how to improve the curriculum.

  • High School Physics Tutor
    • Customize learning materials based on student's proficiency level.
    • Student was accepted by NTU EE

Projects, the big ones

    Building Extraction with Enhanced Boundary
    Designed a semantic segmentation model to extract buildings from aerial images. Utilized building borders to improve the performance and developed a post-processing algorithm to refine the results further.

    Python   TensorFlow   Keras   U-Net   QGIS   Label Studio

     Emotional Voice Converter
    Two models (CNN and CycleGAN) were constructed to convert the emotional features of a given speech. Contained extensive audio signal processing and evaluation. Introduced a fresh data analysis method for examining emotional cues present in audio results.

    Python   TensorFlow   PyTorch   CycleGAN   Audio-processing

     Quantitative Trading Web App
    A Python/Flask web app integrating technical indicators, option-implied volatility forecasting, and LightGBM ML for sentiment scoring - a powerful quantitative analysis toolkit for options traders.

    Python   Flask   LightGBM   Pandas   Option Strategy

Projects, the small ones


Potential stock finder

Similar to stock screeners. But with a better feature: easy customisation.




Web scraping   Pandas   NumPy

Moving avg trading strategy

An easy to follow trading strategy based on Dollar Cost Averaging. Ideal for busy office workers who still want to invest. Includes backtesting and profit visualization.

Matplotlib   Pandas   NumPy

Fake reviews generator

Generated synthetic reviews via large language models. Tested their human likeness via SOTA fake review classifiers.


Sklearn   Transformers   NLP   Pandas   PyTorch   NumPy

Basic information retrieval model

Pre-processing including cleaning, tokenising, lemmatising. Models including BM25, Laplace smoothing, Lidstone correction, and Dirichlet smoothing.


nltk   csv   Information retrieval

Deep learning basics

Coding from scratch to gain familiarity with various deep learning techniques. Including SGD, DenseNet, data augmentation, and cross-validation.

TensorFlow   NumPy   SGD   CV