Last week two of our top engineers and I took a trip to ExCeL, London for Amazon’s Startup Day, an event focused on Startups and innovation ahead of their main Summit event.
Each squad has all of the skills required to take an idea from inception to delivery to their users.
This approach has been embraced widely across software delivery teams, breaking down traditional barriers and siloed thinking that you’d see in large companies in the past.
As your startup grows, it is important to identify the right points at which to break up teams and maintain the agility that helps you to move quickly in response to changing business or market needs.
Data Driven Product Management
Onfido gave a talk on something our own Senior Product Manager, Eddie Sawyers, embraces here at Reward Cloud - the principle of data-driven Product Manager. Base your decisions on hard data whenever possible rather than Vanity Metrics or gut feeling alone.
“Vanity Metrics may help you understand the high level trends in your business, but do little to tell you how to get better.”
Of course, getting to that data can involve a lot of systems...
As a happy customer of Starling Bank myself, I’m always interested in what this ambitious mobile-only challenger bank have to say. This was the second talk I’ve seen from Sam Everington, this time on “Customer Obsession in Retail Banking” and later at the Innovation Hub a talk from Steven Newson, Director of Technology talking about their architectural approach.
If you’re interested in a modern bank and their approach, I recommend taking a look at this Amazon’s Case Study.
There were a couple of talks on Machine Learning and Artificial Intelligence that I found very interesting. Sara Mitchell, AWS Solutions Architect moderated talks from Mirriad, a company that enables seamless in-video advertising using streaming image recognition and Babylon Health, a chatbot to help deal with medical queries using machine learning.
Both talks provided a nice introduction in to the life cycle of all Machine Learning projects;
- Collecting and cleaning your data
- Divide that data in two, a set for training and a set for testing/validating your model
- Train and tweak your model - this is a very iterative process and where the Data Science magic really comes in
- More testing / validation
- Release into the wild…
This slide is from Mirriad’s talk, however what this diagram hides is the continual work to train and tweak the model after it is live, constantly validating that the results produced are giving your business what it needs.
Amazon’s ML services provide a great way to get Machine Learning up and running with minimal initial outlay - watch this space, we have some great ideas to make use of these in the Reward Cloud platform in the future.
Not all serious
Dimitrios and Joe got to meet the AWS Ninja, experience some close-up magic tricks and I won a nano drone for the office (I think we could use some tips on how to fly the drone, though the only casualty so far has been productivity - thankfully the battery life is too short to cause real damage).