Updating Results

Australian Taxation Office (ATO)

  • 1,000 - 50,000 employees

Tim Cottew

6.45 AM

First thing every morning is going out into the backyard and waking up with an exercise routine. I figure if I can get out of bed for chin-ups, I can get out of bed for anything. It’s also partly because I don’t spend much time outside during the week. I skim through the paper over breakfast. I often end up reading for too long, especially if I find articles relating to taxation or other public policy. I find it useful to keep up to date with current affairs to give me perspective on my day-to-day work. It looks like I’m running a bit late, but this isn’t a problem due to flex time and I can make up the time this evening.

8.30 AM

It takes an hour on public transport to get to the office so I think about how I will solve yesterday’s lingering coding problems. At the end of next week, I need to present the findings from an analysis I’ve been working on for about a month. I read sections of textbooks and online resources to help me with a data visualisation problem I’m having. I come across the idea of using violin charts, a happy mixing of box plots and density charts.  

9.30 AM

I take the stairs up to my floor and greet those in the data science team who are in early. I chat with the other data science grad, while my desktop PC starts up. I check my emails first. Most are just internal news and training updates, but I do have a response from my client in the Superannuation business line. Earlier this week I was scrutinising the SQL code (all 2000 lines) used to extract the data for my analysis and noticed a previously overlooked exemption that may affect the results. After team consultation it is recommended, the data remain unchanged. I can go back to further investigation if I finish ahead of schedule.

Day in the life_ATO_Tim Cottew_800x600 2018

I log into our AnalyticsNet, a collection of servers that can handle (just) the stresses our data scientists put under it. I resume my session and try presenting the results as violin charts. Tasks such as this seem simple until I realise a few things need to change before I can even generate graphs. The way the data is structured is currently incompatible. Code is rewritten, data is reloaded and about a third of my time is spent googling error codes and reading Stack Overflow. It can be frustrating work to muck around with data and squash bugs, but eventually I have the satisfaction of getting the graphs looking how I want them to look. I show them to my manager to ask his opinion on how readable they are. He likes them but offers some suggestions in tailoring them for a more general audience.

12.00 PM

I have a meeting with my previous team from rotation 1 – Tax Gap. Every week a volunteer will share a useful package for the programming language R (which is the main language I use) and take everyone through a lesson on how to use it.  This week is “lubridate”, an easier way to interact with date/time values in the base version of R. I put on my headset, dial into the meeting and open up screen share. I look forward to learning new and better ways of doing my work – it’s also a way to keep in touch with my old team. At the end of the lesson I have a chat with my old manager and make plans to go rock climbing on the weekend.

12.30 PM

I head up to the top floor to have lunch with 15 of the other graduates at my site. There is the usual mix of small talk, banter and the occasional, careful, skirting around political topics. It’s a great opportunity to learn how interesting other areas of the tax office are, and what everyone is up to. Sometimes we’ll play cards or some friendly chess.

1:00 PM

Back to coding. I get started on my manager’s suggestions. Thankfully these are relatively easy to implement and I don’t have to look up how to do anything. There is only 15 minutes until my next meeting, no time to really start doing the next section of my analysis, so I prepare for the meeting instead.

1.30 PM

I meet with the Melbourne graduate engagement committee via telepresence. The committee is new this year and gives grads the time, place and finances to manage our own social events. We run through the agenda, taking on responsibility for different tasks such as booking a venue for the end of year party, or giving reports on our own projects. I update the committee on the ATO Movember group – so far we’ve got 16 grads on board and raised about $700. I ask everyone to nominate other grads to participate!

2.00 PM

Back to coding (again). I start the next part of my analysis, drilling down to an employee level. This is slow going as there are millions of rows of data and I’m not ready to cull variables yet. Every small test takes at least a minute to run which really adds up. In about a month’s time I aim to have set up a machine learning algorithm to find the important variables and create a predictive model that the client can use.

3.00 PM

There is a Smarter Data meeting where the entire business line gets together via telepresence while the Senior Executive Service leaders outline their new strategy document and how that fits in to the ATO’s corporate objectives. A few other presentations from senior analysts are given but my mind wanders and I’m really thinking about my code.

4.00 PM

I return to my desk and find some classic dad jokes from a grad recently back from paternity leave, I send him a few instant messages over our intranet and we chat about his kid for a few minutes. I write and send the Movember email to the grads. Then, as usual, back to coding! I continue to explore the lowest level of the data and use my previous aggregate results as a template. I focus intently on just fixing my bugs and pushing the analysis forwards only noticing how late it’s getting when my stomach starts growling.

6.30 PM

I pack up everything on my desk and put it into my locker – I’m not allowed to leave anything out in the open for security reasons and have to shut down my desktop PC. Luckily AnalyticsNet is accessed remotely so it can run code 24/7, not that I need to do that today. I catch one of the infrequent express trains back home; normally I’m too mentally drained to keep doing data science so instead I’ll listen to podcasts. This evening is Dan Carlin’s Hardcore History, Kings of Kings about the Achaemenid Persian Empire, at the time the largest empire in the ancient world (550 to 330 BCE).

7.30 PM

I start cooking dinner and continue listening to podcasts.

8.30 PM

I’m tempted to play computer games but instead just read the news.

9.30 PM

I spend a few hours looking at headphones online; I need some that are suitable for the office as it can get hard to concentrate in an open office environment!

11.00 PM

Bed.