Perplexity Machine Learning Research Engineer Internship 2026: Eligibility, Skills, Location and Apply Details
Perplexity is offering a Machine Learning Research Engineer Internship in its Search department in Berlin. The position is a full-time internship focused on advancing search quality through machine learning models, data, research and engineering.
The internship program lasts 12 to 24 weeks. The supplied listing describes the program as in-person at the Berlin office, while the position metadata identifies the location type as Hybrid. Candidates should therefore review the application details for the exact working arrangement applicable to the internship.
The role is particularly focused on retrieval and ranking models, large-scale deep learning, representation learning and Retrieval-Augmented Generation (RAG). Candidates with strong PyTorch experience and research exposure in areas such as information retrieval, contrastive learning or multimodal modeling may be well suited to the opportunity.
Perplexity Machine Learning Research Engineer Internship: Overview
| Particular | Details |
|---|---|
| Company | Perplexity |
| Position | Machine Learning Research Engineer Intern |
| Department | Search |
| Employment Type | Full Time |
| Location | Berlin |
| Location Type | Hybrid |
| Internship Duration | 12–24 weeks |
| Stated Program Format | In-person in Berlin office |
| Primary Areas | Search, Retrieval, Ranking, Representation Learning |
| Main Framework | PyTorch |
| Distributed Training | PyTorch Distributed, DeepSpeed, FSDP |
| RAG | Required area of work |
| Research Publications | Publication record in AI/ML conferences or workshops |
About the Machine Learning Research Engineer Internship
The internship is centered on improving the quality and capabilities of search systems. Interns will work on machine-learning research and engineering problems involving models, data and search infrastructure.
The role combines large-scale deep-learning engineering with research in representation learning. It also includes work on RAG pipelines designed for grounding and answer generation.
The internship is part of Perplexity’s Berlin internship program, with a duration of 12–24 weeks.
Key Responsibilities
Improve Search Quality
The intern will work to continuously improve search quality using different sources of leverage, including:
- Machine-learning models
- Data
- Tools
- Other approaches that can improve search performance
The position therefore involves both research and practical engineering aimed at measurable search improvements.
Train Large-Scale Deep Learning Models
A major responsibility is training and optimizing large deep-learning models.
The role specifically mentions:
- PyTorch
- PyTorch Distributed
- DeepSpeed
- FSDP
- Hardware acceleration
The primary modeling focus is on retrieval and ranking models.
Research Representation Learning
The intern will conduct research in representation learning, including areas such as:
- Contrastive learning
- Multilingual representation learning
- Evaluation
- Multimodal modeling
- Dense vector representations
- Sparse vector representations
- Representation fusion
- Cross-lingual representation alignment
- Training-data optimization
These techniques are applied in the context of search and information retrieval.
Build RAG Pipelines
The role also involves building and optimizing Retrieval-Augmented Generation (RAG) pipelines.
These pipelines are used for:
- Grounding
- Answer generation
Eligibility for Perplexity Machine Learning Research Engineer Internship
The supplied qualifications indicate that this is a technically and research-oriented machine-learning internship.
Search and Retrieval Knowledge
Candidates should understand search and retrieval systems, including principles and metrics used to evaluate their quality.
Knowledge of how retrieval systems are assessed is therefore an important part of the role.
PyTorch Expertise
Strong proficiency with PyTorch is specifically required.
Candidates should also have experience with:
- Distributed training
- Performance optimization
- Large-model training
The listing identifies PyTorch Distributed, DeepSpeed and FSDP as examples of distributed-training technologies.
Representation Learning
Candidates should be interested in representation-learning techniques, including:
- Contrastive learning
- Dense representations
- Sparse representations
- Representation fusion
- Cross-lingual representation alignment
- Training-data optimization
- Robust model evaluation
Research and Publication Requirement
A notable qualification for this position is a publication record in AI/ML conferences or workshops.
The listing provides examples including:
- NeurIPS
- ICML
- ICLR
- ACL
- EMNLP
- SIGIR
This makes the role particularly relevant to candidates with substantial academic or research experience in machine learning, information retrieval or related fields.
Work Location and Internship Duration
The internship is based in Berlin.
The position metadata lists the location type as Hybrid, while the internship-program description states that the program is full-time and in person in the Berlin office.
The internship duration is 12–24 weeks.
Because the supplied information contains these two work-location descriptions, candidates should confirm the exact working arrangement during the application process.
Salary and Benefits
The supplied job description does not specify a salary, stipend or benefits package for the internship.
No compensation figures should therefore be assumed from the available information.
Selection Process
The supplied vacancy information does not describe the specific selection stages for this internship.
It does not confirm whether candidates will complete a coding assessment, research interview, technical interview, HR interview or other assessment.
Candidates should refer to the application process for any recruitment-stage information provided by Perplexity.
How to Apply
Candidates interested in the Machine Learning Research Engineer Internship can apply through the official Perplexity job listing.
The supplied vacancy provides separate pages for the job Overview and Application.
Applicants should review the qualifications carefully, particularly the requirements around PyTorch, distributed training, search/retrieval systems and research publications, before submitting their application.
Important Points Before Applying
- The position is Machine Learning Research Engineer Intern.
- The department is Search.
- The job location is Berlin.
- The internship is full time.
- The program duration is 12–24 weeks.
- The metadata describes the position as hybrid, while the program description says it is in person at the Berlin office.
- Strong PyTorch proficiency is expected.
- Experience with distributed training and large-model optimization is required.
- The role focuses heavily on retrieval and ranking.
- Representation learning is a major research area.
- RAG pipeline development is part of the role.
- Knowledge of search and retrieval evaluation is expected.
- A publication record in AI/ML conferences or workshops is listed among the qualifications.
- Salary and benefits are not specified in the supplied listing.
Frequently Asked Questions
What is the Perplexity Machine Learning Research Engineer Internship?
It is a full-time internship in Perplexity’s Search department focused on machine-learning research and engineering for search, retrieval, ranking, representation learning and RAG systems.
Where is the internship located?
The position is located in Berlin.
How long is the internship?
The internship program lasts 12 to 24 weeks.
Is the internship remote?
The supplied listing does not describe it as remote. The metadata identifies the position as Hybrid, while the internship-program description states that it is in person in the Berlin office. Candidates should confirm the exact arrangement during the application process.
What machine-learning framework is required?
The qualifications specifically call for strong proficiency with PyTorch.
Does the role involve distributed training?
Yes. The position specifically mentions distributed training technologies including PyTorch Distributed, DeepSpeed and FSDP, along with performance optimization for large models.
What research areas are relevant?
Relevant areas include information retrieval, ranking, representation learning, contrastive learning, multilingual modeling, multimodal modeling and robust evaluation.
Is research publication experience required?
The qualifications list a publication record in AI/ML conferences or workshops, with examples including NeurIPS, ICML, ICLR, ACL, EMNLP and SIGIR.
Does the internship involve RAG?
Yes. Building and optimizing RAG pipelines for grounding and answer generation is one of the listed responsibilities.
Is the internship salary specified?
No. The supplied job description does not provide a salary or stipend figure.