Perplexity Internship — ML Research Engineer, Berlin
- Deadline
- Rolling
- Deadline type
- Rolling
- Region
- Europe
- Category
- Internship
- Eligibility
- Strong ML background with experience in PyTorch, distributed training, and large-scale deep learning. Understanding of search and retrieval systems preferred. In-person Berlin, 12-24 weeks.
- Remote
- No
- Last verified
- Verified on 2026-07-06
💡 ML research internship at Perplexity's Berlin office — search retrieval, ranking, and large-scale model training, 12-24 weeks.
No fixed closing date published. We show this as rolling and exclude it from countdown reminders.
12-24 week full-time ML Research Engineer internship at Perplexity's Berlin office. Focus on pushing search quality forward through models, data, and evaluation tools. Responsibilities include training and optimizing large-scale deep learning models (PyTorch, distributed training, DeepSpeed, FSDP), research in representation learning (contrastive, multilingual, multimodal), and building RAG pipelines for answer generation. Rolling applications — no hard deadline stated.
Frequently asked questions
- Who can apply?
- Strong ML background with experience in PyTorch, distributed training, and large-scale deep learning. Understanding of search and retrieval systems preferred. In-person Berlin, 12-24 weeks.
- Is this remote?
- No, this is not listed as a remote opportunity.
- Is there a hard deadline?
- No — this opportunity has a rolling, ongoing application process with no fixed closing date.
Related opportunities
- NAVER LABS Europe — LLM Research InternshipInternship · Europe
- Machine Learning Internship — Iris.aiInternship · Europe
- BCG X — Forward Deployed AI Scientist Internship (UK)Internship · Europe
- IBM Research Europe — AI Internship (Zurich)Internship · Europe
Check your eligibility
Answer a few quick questions — nothing is saved, this just checks against this opportunity's stated criteria.
Biweekly opportunity digest
This project is no longer being actively updated, so new subscriptions are currently closed. See /about for details.