Posted 1 hour ago
Research engineer
@ turbopuffer
Canada
Overview
turbopuffer is hiring a research engineer to build next-generation embedding and reranking models for search. The role integrates and evaluates models with the inference team, designs and trains models that work closely with the database, and communicates model choices and trade-offs internally and to customers.
Responsibilities:
- Integrate and evaluate embedding and reranking models
- Design, train and evaluate search models
- Communicate model choices and trade-offs
- Work with the inference team
Requirements summary:
- Experience with embedding or reranker model pre-training or fine-tuning
- Experience with large-scale distributed training
- Ability to reason about large datasets
- Ability to explain research to non-specialists
- Experience shipping to production
- Current knowledge of retrieval and search research
Source and dates
Employer career site / ATS feed · ashby
- Employer posting date
- 2026-11-16
- First seen by our source
- 2026-11-16
- Source record updated
- 2026-11-16
These dates describe the source record, not a live verification that the employer is still hiring. Check the original posting.
Job details
| Department | Software development |
| Hours per week | 40 (estimated) |
| Work setting | Office |
| Position | Mostly sitting |
| Physical demands | physical low · computer high |
Summary and details written by AI from the posting; the employer’s posting is the authority.
turbopuffer
Saved company profile
Historical company information from a saved The Jobs API profile (LinkedIn/Crunchbase fields). It is not live-verified; figures may have changed. Missing values remain unknown.
serverless {vector, full-text} search from first principles on object storage. faster, 10x cheaper, extremely scalable
Full company description
turbopuffer is a fast search engine that combines vector and full-text search using object storage, making all your data easily searchable. Using only object storage for state and NVMe SSD with memory cache for compute, turbopuffer scales horizontally to handle billions of documents. The system caches only actively searched data while keeping the rest in low-cost object storage, offering competitive pricing. Cold queries process 1 million vectors in 500ms (p90), while warm queries take just 20ms (p50). This architecture means it's as fast as in-memory search engines when cached, but far cheaper to run. Storing data in cache and object storage costs less than traditional replicated disk systems, even for frequently accessed data.
- LinkedIn followers
- 7,114
- People associated on LinkedIn
- 37
- Business categories
- Information Technology
- Database
- Other recorded names
- turbopuffer
- Profile identifies a recruitment agency
- No
- Reported total investment
- 0
Currency and observation date were not supplied; this is not a current valuation or a funding round.
turbopuffer is a fast search engine that combines vector and full-text search using object storage, making all your data easily searchable. Using only object storage for state and NVMe SSD with memory cache for compute, turbopuffer scales horizontally to handle billions of documents. The system caches only actively searched data while keeping the rest in low-cost object storage, offering competitive pricing. Cold…
| Field | Value |
|---|---|
| Industry | Technology, Information and Internet |
| Company size | 11-50 employees |
| Organization type | Privately Held |
| Year founded | 2023 |
| Open jobs on Jobsy | 6 |
| View page |
turbopuffer is looking for a research engineer who wants to build the next generation models for search.
what you'll do:
work with our inference team to integrate & evaluate the best embedding & reranking models
design, train and evaluate state-of-the-art models that integrate closely with the database
communicate internally & externally about model choices, evaluations and trade-offs
automate yourself out of a job
must have:
research
you’ve worked extensively with embedding/reranker-model pre-training and/or fine-tuning
you have experience on how do to a large scale, distributed training run
can work with and reason about large datasets
you can communicate your research findings to people outside of the field & guide customers in their models choices & questions
you love shipping things to prod
domain expertise
you have an opinionated take on model benchmarks/evaluation
you’re up to date on state-of-the-art retrieval & search papers
you know how to quickly iterate on new research ideas to verify/debunk hypothesis
typist - you can write well and can explain complex ideas simply
human - you build trust and admit what you don’t know
nice to have:
experience with mode inference and managing GPU infra
worked in a systems language (Rust, C++, Java, Go, etc.)
prior experience in vector and full-text search
tpuf is hardcore & whimsical. we value:
overstep > understep
show > tell
complexity is earned
showing up as part of our customer’s team
making stuff easy to react to
Journal
Only you can see this. Kept on this device, and in your account when you sign in.