About the team
The Antifraud team in the Data Science Department develops user scoring algorithms and communication analysis systems. We fight fraud: mass registrations, account takeovers, phishing, and other abuses. The main focus now is the launch of antifraud systems for a new product line based on classic ML and NLP.
At Avito, we create a respectful, safe, and inclusive environment, fostering fair treatment for all employees based on their personal achievements and professionalism.
This vacancy is available for people with disabilities.
What tasks await you
- Develop and improve fraud detection algorithms.
- Create and refine heuristics and ML models.
- Work with text classification tasks: experiment with different architectures and targets.
- Prepare, label, and analyze data.
- Evaluate model quality and track their performance in production.
- Analyze the effectiveness of solutions based on data and business metrics.
- Participate in the implementation of ML solutions.
We expect you to
- Have basic knowledge in ML / Data Science and be familiar with NLP tasks.
- Be able to work with data and write in Python (Pandas, NumPy), be familiar with Jupyter Notebook.
- Master SQL for data extraction.
- Understand basic ML methods and their applicability.
- Be ready to learn, work in a team, and accept feedback.
- Read technical materials in English.
- Be ready for a large amount of verbal communication.
It would be great if you
- Participated in ML competitions (Kaggle and similar).
- Have pet projects or educational projects in ML.
- Have experience working with different types of data (tabular, text, event-based).
Working with us means
- Official paid internship.
- Internship period — 6 months.
- Work format — remote, hybrid (discussed individually based on IPRA), 5/2.
- Great adaptation, professional mentors, and immersion in tasks from day one.
- Intern community for sharing experience and mutual support.
- Equipment for productive work and meal compensation.
- Access to corporate online libraries "MYF" and "Alpina".
