Elastic
Elastic, the Search AI Company, enables everyone to find the answers they need in real time, using all their data, at scale — unleashing the potential of businesses and people. The Elastic Search AI Platform, used by more than 50% of the Fortune 500, brings together the precision of search and the intelligence of AI to enable everyone to accelerate the results that matter. By taking advantage of all structured and unstructured data — securing and protecting private information more effectively — Elastic’s complete, cloud-based solutions for search, security, and observability help organizations deliver on the promise of AI.
The Role
In October 2025, Elastic acquired Jina AI, a team that spent six years building Search Foundation Models. We develop world-class embedding models, rerankers, and retrieval systems—from multilingual to multimodal, from 8K context windows to binary quantization. Our models see 5M+ downloads monthly on HuggingFace and process 200B+ tokens daily through our APIs. We publish at NeurIPS, ICLR, ICML, EMNLP because we believe shipping a model without a paper is shipping without understanding.
We operate as a focused, high-output team of ~20 people. Our structure is flat, our feedback loops are short, and everyone—including leadership—writes code and ships models. We believe that in a fast paced field, extreme focus and efficient execution are what separate good teams from great ones.
We're now building out the Search Foundation Models at Elastic and looking for exceptional researchers to join us.
What You Will Be Doing
- Architectures: Encoder/decoder designs, single vs. multi-vector representations, early vs. late interaction mechanisms
- Scale & Efficiency: Long-context retrieval (8K+), sparse embeddings, low-bit/binary quantization, Matryoshka representation learning
- Modalities: Multilingual models, multimodal/omni retrieval, vision-based document understanding
- Training: Synthetic data generation for hard negative mining, retrieval model distillation, learning directly from embedding space
- Emerging Directions: R1-style reasoning retrieval models, task-specific vs. multi-task training strategies
Additional Information - We Take Care of Our People
As a distributed company, diversity drives our identity. Whether you’re looking to launch a new career or grow an existing one, Elastic is the type of company where you can balance great work with great life. Your age is only a number. It doesn’t matter if you’re just out of college or your children are; we need you for what you can do.
We strive to have parity of benefits across regions and while regulations differ from place to place, we believe taking care of our people is the right thing to do.
- Competitive pay based on the work you do here and not your previous salary
- Health coverage for you and your family in many locations
- Ability to craft your calendar with flexible locations and schedules for many roles
- Generous number of vacation days each year
- Increase your impact - We match up to $2000 (or local currency equivalent) for financial donations and service
- Up to 40 hours each year to use toward volunteer projects you love
- Embracing parenthood with minimum of 16 weeks of parental leave
What You Bring
- PhD or MSc in Computer Science, Machine Learning, NLP, or related field
- 3+ years of experience training production models (full training pipelines, not only fine-tuning)
- 3+ years of professional Python development (PyTorch, Transformers, distributed training)
- Strong foundation in information retrieval and neural search
- Track record of publications at major venues (ACL, EMNLP, NeurIPS, ICML, ICLR, or equivalent)
Bonus Points
- Experience training embedding models or rerankers
- Understanding of token economics: batch sizes, throughput, latency, and context length tradeoffs
- Ability to rapidly prototype—read a paper and have a working experiment within days
- Contributions to open-source ML projects
- Familiarity with retrieval benchmarks and their limitations
