7.30 AM
I wake up around 7.00 am and start my day with a hearty breakfast while watching the news. If the weather is nice like today, I’ll ride my bike into the office (otherwise I take the tram if it’s rainy). I park my bike in the basement parking area and shower at the end-of-trip facilities available at the office.
8:30 AM
I get to my desk around 8.30 am, and spend some time catching up with my colleagues who I hadn’t seen in person for a couple of days (as I was working from home). I then begin my workday properly by checking my emails and calendar. Today there are only a few emails relating to Learning and Development, and some wider organisational updates and newsletters. I only have one major meeting today, and I don’t need to prepare anything.
9:00 AM
I log into my team’s virtual machine and check the results of the model I had sent to be trained overnight. I’m trying to develop a model to understand and classify the content of financial documents, based on deep-learning techniques in Computer Vision and Natural Language Processing. I open the output files and check the predictions generated by the model. I then spend some time evaluating the results of the training and notice a few inconsistencies in the output. I proceed to write down a few fixes I’ll have to implement in the post-processing steps in the code.
10.30 AM
At 10.30 am, we have our daily team stand-up on Microsoft Teams. In this meeting we all go around the (virtual) table and update the team with our progress since yesterday, and if we have questions or are facing any blockers. This is a good opportunity for me to take a step back from the depths of the code and to put what I’m trying to achieve into perspective. It’s also a great chance to see if anyone else is facing similar problems.
10.45 AM
After the stand-up, I spend the remainder of the morning chasing the bugs I’d identified in the outputs throughout the code. This requires me to do a bit of Googling, running some small unit tests, and stepping through each fix one at a time.
12.30 PM
After a productive morning, I stop to have lunch. I really like spending time outdoors, so I often ask a couple of my colleagues to come have lunch in the park across the street from the office. This is a great chance to get some fresh air, chat about non-work-related stuff, and get a change of scenery before heading back inside for the afternoon.
1.15 PM
After lunch, I go back to chasing up the bugs in the code I’ve been working on. I have a brief call with my team leader to chat about a specific problem, and the best approach to tackle it. We discuss our potential options and end up settling on a particular method, which I then proceed to implement. After this, I send the script off to run in the background while I turn my attention to the next meeting.
2.30 PM
This afternoon we have a wider Data Science Community of Practice meeting, which takes place every month or so with the entire Data Science branch. These sessions are a great opportunity to hear about other projects going on at the ATO, which always has several interesting Data Science activities being carried out due to the sheer size and scope of the organisation. Today’s talk is about Time Series forecasting, which I find really interesting as I’m not exposed to this area in my normal work.
3.30 PM
After the Community of Practice session, I take a short coffee break with my colleagues to stretch our legs. We head to the café downstairs and chat about the meeting we just had, as well as our plans for the evening.
3.45 PM
After the much-needed caffeine boost, I check the results of the script I’d sent off to run earlier this afternoon. The bugs I’d identified earlier were happily all working now! This always gives me a great feeling of accomplishment and satisfaction, as well as a clear way to measure progress. Not every bug and feature can be resolved within a few hours, but the feeling you get when there is a breakthrough is always fantastic.
I spend the rest of my workday thinking about the next steps in the model development. I do a bit of research online as to the best practices relating to fine-tuning the model once the main training step is complete. I write some items in a To-Do list for tomorrow – mostly relating to papers I want to read and specific ideas to research further.
4.45 PM
I log off from my computer and head back to the basement to pick up my bike. I then ride home listening to some music and get back home just after 5.00 pm.