The Detection team at HUMAN is the beating heart of our company - finding and stopping bots. We build the algorithms that run behind the scenes analysing and decisioning on our vast number of daily events. We deploy advanced statistical and machine learning models that need to be flexible for different deployment scenarios and robust to a fraudster’s favourite tactics: evasion and deception.
The human we need for this role will work on BotGuard: our product for cybersecurity and account protection. BotGuard provides real-time insight and security for customers fighting account takeover, account creation fraud, payment fraud, and content abuse.
What you will do:
- Devise experimental and production techniques (supervised, unsupervised, statistical, and rule-based) which leverage the signals we collect from browser and in-app interactions to produce robust ‘bot markers’ for our product.
- Gain a working understanding of application / account based threat models to best identify analytical approaches that will work.
- Help the team counter ‘named attacks’, ongoing bot campaigns that we work to eliminate, by understanding how their operators adapt and identifying the weaknesses in their strategies that allow us to keep our markers accurate.
- Be an active participant in team responsibilities, including reviewing other data scientist’s work prior to launching new detection rules/algorithms, our ‘sheriffing’ process where we proactively monitor anomalies arising in our systems and taking part in ‘tech-debt days’ to keep our codebase clean and healthy.
- Collaborate with our product / commercial team to develop new data-intensive product features that will delight our customers.
- Contribute to our growing DS toolsets, ‘creating tau’ by empowering the other humans around you.
Who you are:
- You’ve worked as a data scientist solving large-scale data intensive problems in production systems (we process trillions of events per day, so literacy with large datasets is a must!).
- You can explain the nuances of different modelling approaches and make informed decisions on which to use, whilst also having the engineering awareness of the realities of shipping working code to customers.
- You are proficient with Python and SQL and familiar with related tools, libraries and platforms (e.g. Jupyter, Pandas, NumPy etc).We don’t ask for specific prior domain experience. However, fraud detection requires identifying anomalies and abuse with minimal ground truth and cybersecurity work requires a savvy awareness of how the web works and how bad actors operate.
- Be prepared to tell us why you think your experience sets you up well.
- You’ve demonstrated that you’re self-driven by learning the skills you need to solve problems you’ve tackled in previous roles.
- You almost certainly don’t have all the skills to thwart all the bots right now, so an appetite for growth is key.