The 2018 Melbourne Data Science Conference (MeDaScIn Conference) featured data science career advice from 6 presenters. The conference, held on Wednesday 26 September 2018, was staged at the NAB Arena in Docklands.
Overall, the conference presented some good common sense career advice though there was little “new” material.
This blog post shows notes taken during the presentations.
Scott Wilson: How to Engage Executives and the Boardroom
Scott Wilson, the founder of Wilson AI, presented on “How to Engage Executives and the Boardroom”.
My impression of Scott was that he is about my age (51), professionally presented and an engaging presenter.
Scott’s key points were:
The average age of directors of ASX100 companies is 61.9 years.
Directors natural state is to assess risk and outcomes. There is pressure to maintain the status quo and they have surprisingly have little influence.
When presenting to directors, always focus on the outcome (rather than the method to arrive at the recommendation).
- The CEO (and senior executives) are the most important stakeholders.
- They control the money.
- They manage and allocate resources.
Always follow the money:
- Revenue, spend, cost, impact.
Departments with a reputation for dark arts, “black box”, “black magic” or with no outcomes
EQUALS
candidates for cost cutting or outsourcing.
How to be a hero
1. Engage the CEO and the C-suite (and then the board).
2. Align to their KPIs and company strategy.
3. Follow the money and focus on outcomes.
4. Keep it (the data science) simple.
5. Execute with speed, again and again.
Share the outcomes loud and proud.
Felicity Splatt from PwC
I missed the title of Felicity’s presentation but it was about overcoming “Imposter Syndrome”.
Felicity is a young (to me) Senior Manager at PwC.
I was amused by the topic of her presentation (Imposter Syndrome) given that Felicity has a PhD from Innsbruck University.
Felicity’s key points:
Presented the CRISP-DM methodology.
- Understand the lineage of every component of the data.
- Discussions with data domain experts are often hugely rewarded with gaining key insights.
Advice for data scientists to avoid imposter syndrome.
1. Accept that no one knows everything.
2. Understand your likes and dislikes with data science.
3. Understand the expectations in the role.
4. Remain curious and informed by attending Meetups and conferences.
5. Build a diverse team / support network.
6. Invest in continuous learning.
Eugene Dubossarsky: What data scientists should ask a propsective employer
Dr Eugene Dubossarsky is a director and principal trainer presciient.com.
There is a worringly high level of churn among data scientists, who either move from employer to employer, become freelance consultants or even drop out of data science all together.
Data scientists should learn tips for avoiding unsatisfactory jobs.
For example, the employer is offering a large salary, but…
a. The provided computer is low power (under-speced).
b. The IT department does not support new data science initiatives.
c. Management has vague, “strange” KPIs or even no KPIs.
Avoid employment where the data science reports into IT.
- Managed like IT, as though it is producing “apps”.
Ideally, the data science function sits in the Intelligence function.
- Attends to new opportunities.
What does a good job look like?
- Tangible objectives.
- Shows that the manager understands to role, i.e. manager can help you to progress?
- Neccessary resources, e.g. adequate hardware.
Questions to ask before accepting an offer
1. How will management support to get the required data.
2. How will management help sell the results up the data chain?
3. Where does analytics sit relative to the providers of: Data; Systems; Assets.
Parting words:
- A data literacy revolution is coming!
Sally Grove: Career advice you didn’t learn at school
Back yourself.
It’s OK to ask for help.
Focus on your strengths.
* Only spend enough effort on weaknesses to “plug the holes until you float”.
Don’t get stuck doing things you’re good at but don’t enjoy:
* Train someone else to replace you.
* Automate where possible.
* Have an open conversation with your manager.
Kill your perfectionism:
* Keep it simple.
Take ownership.
Find out the Why:
* Ask follow-up questions.
No one cares how you got there:
* Tell them the insights.
Don’t hide in the corner:
* Have conversations, not emails.
* Say “hi!”.
Be aware of office politics:
* Build a support network.
Get comfortable being uncomfortable:
* Process can be painful but completing brings a great sense of achievement.
Ask yourself “What’s the worst thing that can happen?”.
It’s OK to take a break / sabbatical.
Megan Vassarotti: Department of Premier and Cabinet.
Manages the Victorian open data program.
Most sourced open data is transport data and spatial data.
Common issues with open data
1. Quality of data when the data is by-product of the primary process.
2. Fear of how the data will be interpreted when out of context.
3. Privacy issues.
4. Limited examples of adding value; i.e. “what’s the point?”.
- Government is investing in API gateways.
Final Thoughts
The conference, in hindsight, was more pitched at students and recent graduates, and contrasted sharply from the 2017 conference, which was held in conjunction with Monash University. The 2017 conference featured considerably more speakers as well as breakout sessions.
I noticed as I completed this blog that the conference transitioned into Hackers Helping Melbourne Symposium with the themes:
- Helping those transitioning into data science with advisory talks, and the opportunity to put themselves in the shop window by presenting their work
- Pitch day for the 2018 Melbourne Datathon (the World’s Biggest*). Over 190 teams have had 2 months to analyse the MYKI data and will present their findings.