ML Research Scientist

Svitla Systems, Inc.; Polska; Hybrid; PLN 394,000 - 656,000; Full time

Benefits offered by this job

Flexible workspace

Bonuses for talks/articles

Generous time off

Learning programs

Tech webinars

Job summary

Svitla Systems Inc. seeks an ML Research Scientist for a full-time role, based in Berlin or Western Poland. The team operates in Berlin with hybrid work and requires office visits in Berlin 2-3 times per week.

You will identify top-priority problems at the interface of generative AI and recommendation, and define a long-term research agenda, collaborating across engineering and product leadership. You will mentor scientists, drive architectural debates, and translate SOTA research into scalable

Qualifications

• PhD or equivalent with track record in research mentorship or PhD in CS, Statistics, Math, or Physics.
• Publications in SIGIR, KDD, NeurIPS, or ICLR.
• 5+ years building large-scale recommender systems in AdTech, e-commerce, or social.
• Deep expertise in PyTorch and large-scale model optimization.
• Hands-on with high-concurrency optimization, quantization, and distillation to run large generative models in production.

Responsibilities

• Identify high-leverage problems at the intersection of generative AI and recommendation and define a multi-quarter research agenda.
• Own Generative Retrieval, Deep Ranking, and Re-ranking across a data stack of hundreds of millions of users.
• Translate state-of-the-art research into ship-ready systems and review architectural decisions.
• Mentor senior and mid-level scientists and drive long-horizon ML investments without a fixed roadmap.

Skills

PhD depth

Publications in top venues

Recommender systems

PyTorch expertise

Model optimization

Distributed systems

Education

PhD in CS/Stats/Math/Physics

Tools

PyTorch

Quantization

Distillation

Job description

Svitla Systems Inc. is looking for an ML Research Scientist for a full-time position (40 hours per week) in Berlin or in Western Poland. The team is based in Berlin - hybrid work; office visits 2-3 times/week IN BERLIN required.

Requirements

• PhD or equivalent research depth, with a track record of technical mentorship/ PhD in CS, Statistics, Math, or Physics


• Publications in SIGIR, KDD, NeurIPS, or ICLR


• 5+ years building large-scale recommender systems in AdTech, e-commerce, or social.


• Deep expertise in PyTorch and large-scale model optimization


• Hands-on with high-concurrency optimization, quantization, and distillation - getting large generative models to run under extreme production constraints, not just in research environments.

Responsibilities

• Identify the highest-leverage problems at the intersection of generative AI and recommendation, define the multi-quarter research agenda, and align engineering and product leadership


• Own how Generative Retrieval, Deep Ranking, and Re-ranking evolve across a stack built on 770 million users' behavioral data.


• Translate SOTA research into systems that can ship, review architectural decisions across teams, and raise the ML quality bar company-wide


• Mentor senior and mid-level scientists, lead architectural debates, and drive long-horizon investments without needing a roadmap handed to you.

WE OFFER

• US and EU projects based on advanced technologies.


• Competitive compensation based on skills and experience.


• Regular performance appraisals to support your growth.


• Flexibility in workspace, either remote or our welcoming office.


• Bonuses for article writing, public talks, and other activities.


• Generous time off, including vacation, national holidays, sick leaves, and family days.


• Personalized learning programs tailored to your interests and skill development.


• Free tech webinars and meetups organized by Svitla.


• Regular corporate online activities.


• Awesome team and a friendly, supportive community!