This public github is one of the core artifacts from my "learning in public" journey into the world of Artificial Intelligence (AI).
I plan to go both deep and wide during this journey in the strange world of AI. By "deep", I intend to cover advanced papers about specific techniques or model architectures that marked significant leaps forward in the state-of-the-art. I also plan to write PyTorch (and maybe Tensorflow too) code to showcase some concepts and also cement my understanding of specific techniques and architectures.
By "wide", I mean that I will cover different modalities like text, image, video, and probably sound too. While my focus will predominantly be on large language models and the Transformer architecture, I plan to explore good-old Convolutional Neural Networks and other model architectures too.
I am Eddie, a software engineer by trade with professional experience in Product Management and of course massive interests in AI. I've worked for large organisations and small ones alike. I've built high frequency trading platforms during my time in Finance, large scale distributed/microservices systems that can handle millions of requests globally during, predictive analytics systems for energy companies.
More recently, I am helping a Fintech company in their digital transformation journey by heading their data engineering and data-science initiatives, leading them into a brave and exciting new world where AI can give them an edge in the marketplace.
I gained some experience in AI a few years ago by undertaking the Udacity Self-Driving Car Nanodegree where I led my team during the capstone project, delivering functional software that autononously drove Udacity's self-driving car in a private track. The software that we wrote could recognise traffic lights, stop the car when necessary, and get the car to move at specific speeds along specific waypoints to reach a given destination. This was lots of fun.
Additionally, I studied many of the modules of the excellent Fast AI course from Jeremy Howard and Rachel Thomas. I also completed Udacity's Computer Vision course and watched many of Stanford's AI lessons. During this period I shared some of learnings through Medium blog posts. My most famous post to this date is a of the SSD Multibox object detection network for Computer Vision.
As you can see, I was very into AI at the time so I also spent some time reading and trying to understand many papers about the field a few years ago.
All the above gave me a good intuitive understanding of neural networks at the time. But a lot of this understanding is rusty. There's also been a lot of new developments. The Transformer architecture is now law in the land the AI. Large Language Models like GPT are reigning supreme and they have taken the world by storm. Their adoption over the last year has been nothing but remarkable. And I feel we are just scratching the surface!
My curiosity and excitement about AI have grown again so much in the last year that I cannot not dive into this field again. I've decided that I want to be part of the movement composed of those who want to learn more about AI, for their own personal satisfaction, and to wield this technology to build cool, useful, and profitable applications and companies in the future.
Thank you for following this journey I am on. You can also connect with me on Twitter and LinkedIn.
LFG 🚀.