AI Research Engineer

LanceDB; United States; On-site; USD 160,000 - 210,000; Full time

Job summary

LanceDB is seeking an AI Research Engineer to join the research team, focusing on end-to-end training flows across different AI domains, and showcasing how LanceDB accelerates research pipelines. This role emphasizes impact, experimentation, and contributing to the product with a researcher's perspective.

You will publish models and research papers, collaborate closely with engineering and product, and help raise awareness of workflow features like distributed indexing and blob-based data

Qualifications

• 5+ years of experience in training deep learning models, including video/world models.
• Proven track record of training SOTA models in industry.
• Strong OSS contributions and maintenance experience.
• Experience with PyTorch, distributed training, and tensor parallelism.
• Ability to map feedback to deliverables and prioritize work.

Responsibilities

• Demonstrate end-to-end training flows across domains.
• Publish models and papers on LanceDB blog and conferences.
• Collaborate with engineering and product for researcher perspective.
• Benchmark and compare training workflows across industry verticals.
• Build and maintain OSS repos and tooling.

Skills

Deep learning

PyTorch

Distributed training

Video models

World models

Open source

Research publication

Tools

GitHub

Docker

Tensor Parallelism

Job description

About LanceDB AI advances at the speed of its research, and research moves at the speed of its data. LanceDB is the AI-native Multimodal Lakehouse: one system where a researcher curates petabytes of video, audio, and every signal derived from them with a few lines of Python, and the next training run starts as fast as the next idea. Customers like Runway, Midjourney, and Netflix build the future of AI on LanceDB, from frontier and world models to robots and autonomous vehicles.

AI advances at the speed of its research, and research moves at the speed of its data. LanceDB is the AI-native Multimodal Lakehouse: one system where a researcher curates petabytes of video, audio, and every signal derived from them with a few lines of Python, and the next training run starts as fast as the next idea. Customers like Runway, Midjourney, and Netflix build the future of AI on LanceDB, from frontier and world models to robots and autonomous vehicles.

About The Role As the AI research engineer at LanceDB, you'll work with the research team to perform fairly open ended research, focused on end to end training flows across different AI domains, showcasing how LanceDB can be used to accelerate research flows. This is an opportunity to pursue your research interest as an engineer, and have a meaningful impact on raising awareness and significantly improve the product.

What You'll Do

• Show how Lancedb can be used for training models end to end from curation to modeling across industry verticals
• Compare the Lancedb stacked workflow with existing standard training flows with well designed and replicable experiments, that may include benchmarking
• Provide core content and work cross-function with to increase awareness for workflow specific features like blobv2, distributed indexing etc.
• Publish models and research papers on LanceDB blog platform, social media, and in AI conferences
• Partner closely with engineering and product to provide feedback from a researcher’s perspective

What We're Looking For

• 5+ years of experience in training deep learning models, not limited to LLM, ideally have worked with video, action, world models before
• Proven track record of training SOTA models in an industry vertical
• Strong experience in building and maintaining popular OSS repos.
• Demonstrated ability to map user feedback from noise to key deliverables
• Excellent prioritization skills and demonstrate execution efficiency
• Strong sense of product GTM, demonstrate ability to balance strategic thinking with hands-on execution
• Passion for staying up-to-date with SOTA AI research and trends

Nice to Have

• Experience with training transformer based models, and post-training/alignment
• Hands on experience with PyTorch, distributed training, and tensor parallelism
• 5+ years of experience, including working at startups