Data Scientist, Help Experience

Vinted group.; Berlin; Hybrid; EUR 80,000 - 120,000; Full time

Benefits offered by this job

Share options

30 days annual leave

EAP

Home office budget €540

Lunch benefit

Team-building events

Monthly shopping budget

Discounted gym membership

Pension Plan with 150% matching

Private health insurance

Life and Disability insurance

Deutschlandticket

Dog-friendly office

Job summary

Vinted Group is hiring a Senior Data Scientist for the Ember team within Help Experience. You will shape the next generation of AI-powered support, building conversational assistants and automated resolution systems to reduce manual tickets and costs.

The role emphasizes multilingual capabilities, production-grade ML deployment, and close collaboration with product, backend, and ML platform teams to deliver scalable solutions for millions of users globally.

Qualifications

• Strong background in machine learning and practical ML deployment.
• Ability to read research and translate to code.
• Fluent in English (written and spoken).

Responsibilities

• Build Conversational AI & Agentic Systems: design, develop, and scale LLM-powered virtual assistants and agents for multi-turn dialogue across languages.
• Drive Autonomous Resolution & Decisioning: create models to autonomously resolve or hand off to humans as needed.
• Leverage GenAI & RAG: implement embeddings, prompt engineering, and fine-tuning for context-aware answers.

Skills

Python

SQL

ML algorithms

Negotiation/communication

Job description

Our mission is to make second-hand the first choice, and we're looking for people who want to help us get there. Every day, we work together to help our members buy and sell pre-loved clothing and lifestyle items, giving each piece a second life – or even a third.
The Vinted Group is made up of three business units that support this mission:
Vinted Marketplace is Europe’s leading platform for second-hand fashion and a go-to destination for all kinds of pre-loved items, with a growing range of categories. Our platform connects millions of members across 20+ markets, helping great items find a new life.
Vinted Go enhances the shipping experience with a vast network of over 500,000 pick-up and drop-off points, partnering with more than 60 carriers across Europe, with added services like item verification for peace of mind on high-value pieces.
Vinted Pay is the newest part of the Vinted Group, dedicated to bringing secure, reliable payments to buyers and sellers across Europe. Seamlessly integrated into the Vinted app, it helps keep every transaction safe, efficient, and easy for our members.
Founded in 2008 in Lithuania, Vinted began as a way for friends to find new homes for clothes they no longer needed. In 2019, we became Lithuania's first unicorn! Today, our headquarters remain in Vilnius, and we've grown with offices across Europe, supported by a team of over 2,000 people.

Information about the position

Help Experience (HEX) supports tens of millions of Vinted members across 20+ markets. When members encounter an issue—from shipping delays to payment questions—we want to resolve it as fast, effortlessly, and accurately as possible.

We are looking for a Senior Data Scientist to join the Ember team within the Help Experience domain.

Ember is building the next generation of AI-powered support at Vinted. Our mission is to serve scalable support automation that users appreciate. We build and scale our conversational Virtual Assistant and automated resolution systems, transforming support from traditional ticketing into instant, conversational problem-solving. By resolving issues autonomously and helping members get back to buying and selling with confidence, Ember directly reduces operating costs and manual ticket volume supporting Vinted’s mission to make second-hand the first choice worldwide.

In this position, you’ll

• Build Conversational AI & Agentic Systems: Design, develop, and scale LLM-powered virtual assistants and agents capable of multi-turn dialogue, understanding nuanced member intent, and reasoning over complex marketplace policies and real-time transaction data.
• Drive Autonomous Resolution & Decisioning: Build decision systems and classification models that know when an issue can be safely and autonomously resolved (including executing automated actions like cancellations, tracking updates, or refunds) versus when to seamlessly hand off to human support specialists.
• Leverage Modern GenAI & RAG Architectures: Implement and optimize Retrieval-Augmented Generation (RAG), embeddings, prompt engineering, and fine-tuning techniques to deliver accurate, context-aware answers across multiple languages and noisy conversational inputs.
• Establish Robust Evaluation & Guardrails: Design offline and online LLM evaluation frameworks (measuring policy compliance, hallucination mitigation, response safety, and containment quality) to ensure automated interactions are dependable and aligned with brand guidelines.
• Ship to Production at Scale: Partner closely with Backend Engineers and ML Platform teams to deploy and serve models in low-latency, high-throughput production environments, monitoring for drift and performance degradation in real time.
• Partner for Impact: Collaborate with Product Managers, Decision Scientists, and Customer Support Operations to identify high-impact automation opportunities, design rigorous A/B experiments, and directly move core metrics like Automation Rate, Containment, and Chatbot CSAT.

About you

• Pragmatic Problem Solver: You have a strong technical background and hands-on experience developing machine learning algorithms. You prefer an iterative approach and can read a research paper and translate it into clean, replicable code.
• Service Mindset: You understand that behind every data point is a real member trying to get help. You succeed when the teams using your tools can make better decisions for our members.
• Technical Expertise: You are highly skilled in Python and SQL. You write clean, maintainable, and structured code that serves as an inspiration for others, valuing reproducibility, clarity, and performance.
• Clear Communicator: You can persuasively explain complex technical concepts and their business implications to non-data stakeholders. You understand that stakeholders need to know why a model gave a certain score.
• Fluent in written and spoken English.
• The opportunity to benefit from our share options programme
• 30 days of paid annual leave
• Confidential Employee Assistance Program (EAP) for you and your family
• Home office support: we provide IT workstation equipment and a personal budget of up to €540 for home workplace furniture
• Lunch benefit per your workday
• Frequent team-building events
• A personal monthly budget for shopping on Vinted
• Access to a discounted gym membership plan
• Pension Plan with Vinted matching 150% of your chosen contribution
• Supplemental Private Health Insurance
• Life and Disability insurance
• A subsidised Deutschlandticket for your commute to the office by public transport
• A dog-friendly office

Working at Vinted

Workation policy

Better balance holidays with workdays by working remotely! Up to 90 days per year in the EU, of these, 21 days can be spent globally. For non-EU citizens, it's 21 days worldwide. This can be combined with time off for vacation or personal time.

Individual learning budget

Each year, you’ll be given a learning budget (starting at €3,000), and a total of up to 10 working days over a 2-year period to support your personal and professional development.

Our hybrid model, with 2 recommended office days a week, gives you and your team the flexibility to decide if and when you want to work from home, and when to catch up in person.

Equal opportunity

We welcome applications from everybody, regardless of your background, identity, or life experiences. Job openings come with guides, not checklists. If you’re excited about a role, but don’t identify with every point in the ‘About you’ section, apply anyway – you might still be the perfect match!

The annual gross salary range for this position is: