Hi,
I am Lenwyn!

Data Engineer and Scientist

Contact

About

I am Lenwyn,

Passionate Data Engineer and Scientist with experience in Python, Machine Learning, Deep Learning, Data Warehousing and Lakehouse, across domains such as Medical and Aerospace.


CV, Click Below!


Skills

My strongest skill till now is working with 3D CT Images, it's generation, processing, annotation and then using the data for training deep learning models.
Inferencing these models and gaining insights while testing against unseen data.
I have experience with textual and numerical data as well, I have previously worked on Twitter data in some of my academic projects.


Python | R | Java | PHP | Perl | JavaScript | C++ | C
SQL | MySQL | PostgreSQL | Athena
Machine Learning | Deep Learning | Image Processing | Data Visualisation | NLP
PyTorch | CNN | U-Nets | ETL | NLTK | Apache Spark
Confluence | Docker | Azure | Kanban | MS Office | Git
3D Slicer | MITK | Plotly | Dash | Matplotlib | Seaborn

Work

Porosity Detection in 3D CT Data of Manufactured Parts:
Anomaly Detection

(Deep Learning, Python, PyTorch, 3D Slicer, SimpleITK)

Successfully developed an U-Net model for detecting porous regions within CT images of complex aerospace parts.
Data Annotation was facilitated using 3D Slicer and MONAI Label.
Model had an accuracy of 86% to correctly detect pores.

Classification of Eye Disease OCT Images using ResNets:
Image Classification


(Deep Learning, Python, Tkinter)

Trained a ResNet model to classify four eye diseases using OCT data of the retina.
Hyperparameter tuning was required to adapt the model to perform better.
Image processing techniques were involved to extract import features.

Sarcasm Detection in Tweets:
NLP

(Python, NLTK, TextBlob, Matplotlib, Sklearn, numpy)

This python based project detects sarcasm in tweets using NLP techniques.
The final sarcasm scores are obtained by aggregating the text score and then using Random Forest classification to predict the class.

Transcription and Summarisation of Lecture Videos using Kaldi in English and German:
NLP, NLG

(Python, Kaldi, Transformers, VoskAPI)

Developed a pipeline which takes video files as an input, first transcribes the video into sub-titles or text files and then summarises the text generated.
This was done in mind to help deaf or hearing disability students during lecture settings.

Comparing the tweets of people during early COVID-19 to later years using Sarcasm Detection:
NLP

(Python, NLP)

Had an interesting idea to compare the tweets of users tweeting when COVID-19 was at its peak to late 2022.
The hypothesis was that people have become more prone to be more "chill" about the pandemic as time passes.
A Sarcasm detection model and Word Clouds were used to compare the data.

Contact

Located: Hamburg, Germany

Phone: +49 1787623208

Email: lenwyn98@gmail.com