Research Staff

About the role

Mixedbread is focused on removing the biggest search bottleneck in AI by building the next wave of information retrieval technology. The team is working toward retrieval systems that are multimodal, transformer-first, and capable of finding the most relevant context across different kinds of content.

This position sits inside a full-stack retrieval research lab that is reimagining search for the era of large models. The research scope includes late interaction approaches, multimodal retrievers, and search pipelines enhanced by LLMs. The work is intended to move the product, Omni, forward while also contributing to the wider information retrieval community.

What you will do

• Take ownership of applied research efforts in information retrieval and contribute hands-on to ongoing projects.
• Improve Omni, the company’s main search platform, with a focus on better retrieval quality and system efficiency.
• Create and test new techniques for late interaction and multimodal retrieval.
• Study LLM-assisted search workflows and assess where they help and where they fall short.
• Work closely with engineering and product partners so research outcomes translate into real product value.
• Share findings through research papers, technical blog posts, and conference presentations.
• Help define the research direction and influence the retrieval roadmap.

Research focus areas

• Late interaction: developing more granular retrieval models that go beyond ColBERT-style approaches.
• Omni-modality: creating retrievers that can work across text, images, and other formats.
• LLM-augmented search: examining how large language models create, rank, and consume search results.
• Evaluation: building new evaluation methods that better match real-world, agent-driven usage.
• Data work: strengthening retrieval signals and improving dataset quality to raise system performance.
• Interpretability: understanding why search systems succeed, fail, or behave unexpectedly.

Requirements

The role calls for experience in information retrieval, embeddings, or a closely related area. A PhD, Master’s degree, or equivalent evidence of high-impact research work is expected. You should be comfortable coding in Python and able to build models from the ground up. A strong interest in moving beyond single-vector similarity methods is important, along with a broad grasp of the machine learning lifecycle, including algorithms, training, data, and efficiency. Clear written and spoken communication is also needed.

Nice to have

• Publications in leading conferences such as NeurIPS, ICML, SIGIR, ACL, or similar venues.
• Open-source work in IR, NLP, or machine learning libraries and frameworks.
• Exposure to late interaction models, transformers, or vector search systems.
• Experience planning and running large-scale training experiments.

Benefits and compensation

• Competitive pay along with equity.
• Health, dental, and vision coverage.
• Visa sponsorship and relocation assistance.
• A budget for professional growth and learning.
• Access to leading AI tools and paid subscriptions.
• Team off-sites and support for conference attendance.
• Transportation assistance.
• Wellness support, including gym membership and sports club subscriptions.
• Food support.

Equal opportunity statement

Mixedbread is committed to maintaining an inclusive workplace and welcomes applicants from all backgrounds. Hiring decisions are made without regard to race, religion, color, national origin, gender, sexual orientation, age, marital status, veteran status, or disability.

Application details

Applicants are asked to submit a resume and provide optional profile links for LinkedIn, GitHub, and X, along with written responses about a project or achievement they are proud of and what attracts them to Mixedbread.