It's been a while since I added a post to this blog, a long while but I was busy ;)
I joined Tesco in May '17 after doing an amazing Cyber Security training for few months. Tesco shockingly was extremely well organised and modern. I was exposed to challenges I never faced before, like up to 100k requests per second on live system, and big installations of Cassandra and Kafka. My team and I also did a bit of caching for tills so they can work for customers even if there is a short issue with internet connectivity.
Then moved from Tesco to Tantalum Corporation in March '18. Did some fantastic telematics on steroids (convenient services for car owners on top of telematics) there - also learned a lot from two extremely skilled professionals working there - especially how to properly use Spring Cloud, Consul, Vault (thanks Hari and Jim). How to setup modern and robust CICD pipeline so it is helping development instead of being an obstacle and many more useful skills.
After Tantalum I joined Oodle Car Finance. When I joined they were already a fairly established company with a system to do credit searches, scorecard, underwriting and fraud checks. I worked with one of a kind CTO (thanks Alistair) who built quite a lot of this solution himself while he was a CTO hiring lots and lots of people along the way - quite a challenge! So my task was to stabilise the system and do some improvements. After about 3 months I knew enough to figure out the cost of implementing all the features we needed. It was high and the risk of still not having a perfect solution was there as well.
We have decided to build brand new solution, perfectly aligned to the company's aspirations - especially on data science front - and regulatory requirements. I have to admit - building a new solution when you have an old one with lots of problems already solved is much simpler than starting from scratch. Then the usual approach was taken, feasibility study, analysis and design. PoC was done in Feb-Mar '20 while hiring my own team, first lines committed in the beginning of Apr and then getting a fully working solution by July, testing it from functional, performance, saturation, load perspectives and with all green and happy, releasing it to production on 11th of August. This was a wild ride without any serious issues whatsoever!!! Because Data Science team was able to develop and test their models and decision engine without the need to change or deploy any code by Engineering team the pace was immense. They were sometimes testing 20-30 new models in shadow mode to find the best performing one and then they were switching our production models in minutes to benefit from the better model. I worked very closely with Marina, she has tons of experience in financial markets and especially lending, underwriting, risk and pricing - I think we were complementing each other nicely which allowed us to move fast and deliver great value. When Alistair and Marina left to chase bigger and better challenges it was time for me to leave as well...