Why did I give up my academia career after Ph.D. and build a start-up?
I just got my Ph.D. in Electrical Engineering in 2022 from Nanyang Technological University and Denmark Technical University, with 22 research papers published in 4 years.
I also received a couple of faculty position offers from universities worldwide. But I was hesitant then.
My Ph.D. journey was tough, like a truly "permanent head damage". It was hard because:
- The knowledge gap in AI was huge for me.
My Ph.D. topic was to research How Artificial Intelligence can be applied in the electric power industry. I was trained to be an electrical engineer with no computer science or artificial intelligence background. Thus, learning from scratch about AI and adopting the methodology to real use cases in a brand new problem, was hard. I spent the entire 2 years finishing my first research work and publishing the 1st journey paper.
- The data collection procedure was painful.
As you know, AI models require data for training. To build a surrogate model for the circuit, if using the conventional approach, I needed to collect about 10k data points while 1 data point means 1 set of circuit experiments (taking about 30min). This was a very time-consuming process.
As a Ph.D. candidate, what I have was TIME. Thus I had 4 years on this.
Imagine this happens to a company, like a startup or an MSE which is looking for high growth. What if they want to develop their custom AI model to empower their business,
Do they have enough talents to bridge the knowledge gap?
Do they have enough resources for data collection and model training?**
As reported, to build the ChatGPT model, more than 3 billion US dollars have been spent on it.
Should AI development be always such a luxury?
I wanted to make a change. I believe entrepreneurship was the bridge to bring technology from the lab to real products and people's real lives.
With Ailiverse, we are holding a vision to make deep learning accessible to ALL. With our award-winning breakthrough in domain adaptation along with few shot learning, Ailiverse NeuCore can make computer vision development easier than ever:
- Only 10% of the training data is required.
- Training can speed up to 10X.
- No expertise in AI is required to build a custom model.
With this, you can build your computer vision models in Minutes!
We are just launching our Beta product, looking forward to any kind of feedback!
PS: Ailiverse has recently been featured by Sequoia India and Southeastern Asia Spark Program, which is an important recognition of our vision!