Company Specific Flashcards
How would you clean data?
- EDA
- Normalization
- Missing data (impute, replace, remove)
- Unify formats
- Deduplication
What kinds of datasets/variables have you worked with?
- Large financial datasets
- Dirty data
s
When was a time where a group had one idea, and you convinced them otherwise?
- Data Strategy
- Agile Scrum
- Team Health check
Tell me about yourself.
–
How do other people around judge you?
- How I treat people
- Work ethic
What is your weakness?
‘Done is better than perfect’.
- imposter syndrome
Why analytics?
–
Why deloitte?
–
Talk about 2 of the biggest challenges that Deloitte is facing.
- Relentless pace of tech innovation (Simon Murphy, Silicon Republic) - placing bets on what technologies will be front and centre in the future
- Serving clients in industries with aging business models - more competition, leaner, customer expectations
- Talent
- Responsible AI
- Data Strategy
- GDPR, Open Banking – consortium, federated data, homomorphic encryption
Where do you see yourself in 5 years?
- People management
- Acquired experience across industries
- Polymath
- Leader
Examples of Deloitte innovation
- TAXIE - Tax division RPA
Krawler
- FirstSense
- Deep web, dark web, darknet crawling platform
- Scrapes stolen credit card information
- Matches partial cards to customer database
- Saved millions in $
Rule Evaluation Framework
- Saves $200,000 per client evaluation
- Developed using Python and R
- Simulating performance of rules on real customer data to see what benefit it could bring clients in production
- optimizing thresholds across a number of metrics - recall, precision, false positive rate, false positive amount, cost ratio, alert rate
- catered for many rule types (list-based rules, regex rules, threshold rules, rules with profiles)
Automated reporting framework
- Python data validation, pulling
- R ggplot2
- Docker
Leadership examples
- Foroige volunteering - Big Brother Big Sister programme
- CoderDojo
- YoPro
- Mentoring at work
- Agile Scrum Advocates group
- Kanban Master
- (Anaplan) - Taken the lead to improve interview processes, speak at recruitment events, first day at women in technology fair
Challenges
- Communicating poor model results back to interested senior leadership
Fraud Modelling
Thirty party vendor partnership
Interactive tool for modelling
Sharing compute resources
Increased recall by 40%, precision by 5%
Now Python based modelling, customized decision tree, neural networks, building an ML pipeline using DAGs to set off execution events
Most important learning
Data Science should be treated with same reverence as software engineering
Strengths
- Curiosity
- Adaptability
- Humility
- Push myself in and out of work
