AI labs – a club for #AI research and a chance to gain hands-on experience with AI


We have been working on this idea over the summer and have now launched the next stage of the AI labs in London

Here are some more details.

Think of AI labs – as a club for AI research

AI labs addresses three problems

a)  Today, even if you are working on Machine Learning (rare it itself) – most people / companies are not working on Deep Learning and AI – even if they are working on Machine Learning. The lab offers you this opportunity to build AI applications and deploy in open source – thus demonstrating your expertise

b)  We will work on specific Robotics / AI deep learning technologies often to implement research papers. This is another big gap in the industry. It’s hard to take a paper and implement it on your own (especially if it’s not really a part of your day job)

c)  We will also work with AI models in the Cloud and Automated Machine Learning – both of which I believe will significantly broaden the market for AI


How we work
a) We work in sprints where you implement code and publish it in Apache v2 license. Thus, you get to demonstrate your technical expertise

b) We typically implement papers around GANs, Reinforcement Learning etc

The overall idea is to gain real experience and demonstrate it in rapid iterations using Agile/Sprint

Steps for the sprints are
a)      We present a challenge/ problem (typically code based)
b)      You choose which sprints you join
c)      The group will solve it in a sprint in collaboration in a timeframe
d)      At the end of the sprint time we provide the solution
e)      The Group then lists the best contributions i.e. who made the best efforts
f)       And if it’s a good effort we publish on an external GitHub listing the contributors

This helps you to create a real GitHub portfolio but also work with others in a collaboration.

We start off with three problems(sprints) initially

You can join more than one sprint.
The three initial sprints are likely to be

a)      Reinforcement learning
b)      NLP (Robotics)
c)      A research paper

Problems are set typically by the tech architect team which includes

Ajit Jaokar

Dr Daria Shamrai

Dr Saed Hussain

Cheuk Ting Ho

Dr Amita Kapoor


Sprints are managed by

Rama Govardhana and Dr Ahmad Abd Rabuh


Examples of technologies / papers
a)  Paper – Trajectory Optimization using Reinforcement Learning for Map
b)  The TRFL library open sourced by Deepmind
c)  Unity 3d and Deepmind 
d)  In future, AWS and Azure AI and some form of Automatic ML


Why Robotics/ Devices?
a) It’s easier to learn AI in a physical context and that knowledge is transferable ex you can apply time series to fintech
b) AI and IoT will be huge but still a niche(hence a gap)
c) It reflects my personal network including existing collaborations like Dobot and my teaching


We also pleased to collaborate with

Barend Botha (Signacore)

Devrim Sonmez

Dobot – for Robotics

eOffice (Pier Paolo and Oscar Chu Ortega).

Many thanks to Barend Botha for the Logo.

If you want to join the Lab and are interested to know more – please connect with me on Ajit Jaokar Linkedin (with a message in the connection request re Lab).  

We are still piloting this initiative so we are keeping the group small so we can learn.

We start small and grow. I am a big believer in AI and invest a lot into it. So, this is a long-term vision.