Data Scientist - Fraud Detection

About Kasada

Kasada is dedicated to eliminating malicious bot activity, protecting millions of online users globally every day. Originating as a small team beneath Sydney Harbour Bridge, Kasada now defends some of the world's largest brands from automated attacks that cause significant financial and operational damage. Our mission is to restore internet trust by stopping automated threats at their first interaction, including novel attack methods.

Role Overview

We are seeking a driven Data Scientist to join our Account Intelligence team, focusing on combating fraud and hostile automation. In this pioneering role, you will develop predictive models and analytical techniques to differentiate genuine users from adversaries, collaborating closely with engineering, research, and security teams to deploy robust defenses. Your tasks will span the entire model lifecycle, from data exploration and model building to performance evaluation and adaptation to evolving threats.

Key Responsibilities

• Develop and refine predictive defenses that mitigate risk while enhancing user experience at Kasada scale.
• Conduct thorough evaluations of model performance and decision-making to ensure robustness against scrutiny from technical and security teams.
• Collaborate with engineers, researchers, and product managers to transition models from development environments into production, iterating based on customer needs and attack trends.
• Ensure models and their outputs are interpretable and transparent for both technical and non-technical stakeholders.
• Analyze extensive datasets to detect anomalies, adversarial behaviors, and emerging attack vectors, incorporating contemporary adversarial machine learning and cybersecurity methods.

Candidate Profile

• Deep curiosity about fraud and adversarial tactics, coupled with a thoughtful approach to the impact of detection models on legitimate users.
• Experience partnering across multidisciplinary teams and effectively communicating complex model insights in accessible terms.
• A minimum of two years’ professional experience in data science or applied machine learning, proficient with statistical methods, time-series analysis, and predictive modeling techniques including gradient boosted trees, random forests, and deep learning.
• Strong programming skills in Python and SQL and familiarity with common machine learning libraries such as scikit-learn, PyTorch, and TensorFlow.
• Hands-on experience deploying and evaluating models in live or simulated production settings, with knowledge of precision, recall, calibration, and adaptive threat responses.
• Exceptional analytical and problem-solving abilities with rigorous quality control standards on your work.
• Comfortable working within cloud platforms like AWS.

Preferred Qualifications

• Previous involvement in fraud detection, trust and safety, account takeover prevention, or abuse analytics.
• Experience building models in adversarial environments where attackers attempt to circumvent defenses.

Technologies Used

• AWS, Clickhouse
• Python, scikit-learn, PyTorch, TensorFlow

Company Benefits

• Opportunity to own equity or stock options, sharing in Kasada’s global achievements.
• Comprehensive family support, including generous parental leave programs and resources.
• Access to wellness programs and confidential Employee Assistance Program counseling for employees and their families.
• Additional leave options including birthday and wellness days.
• Annual company retreats for collaboration and team bonding.
• A pet-friendly headquarters located in Sydney.

Additional Information

Kasada encourages applications from people of all backgrounds and experiences and promotes a supportive interview process with accommodations available as needed. Candidates are invited to disclose preferred pronouns and any interview access requirements. Note that Kasada participates in E-Verify for U.S.-based applicants.