ApplyBlast uses AI to match you with the right jobs, tailor your resume and cover letter, and apply automatically so you can land your dream job faster.

Features

Auto Apply Resume Optimizer Cover Letter Optimizer Interview Prep

Company

Social Media

Need Help?

team@applyblast.com

Research Scientist – Diffusion Models at Granica - ApplyBlast

Research Scientist – Diffusion Models

Granica

3 months ago

Onsite

Full Time

Senior

0 applicants

View Job Listing

Apply to 100+ jobs

About this role

As a Research Scientist or Ph.D. researcher at Granica, you will develop novel diffusion models and generative learning algorithms specifically tailored for Large Tabular Models. This role focuses on creating efficient training methods and representation learning techniques for structured enterprise data while collaborating with Stanford Professor Andrea Montanari. You will design rigorous experiments to evaluate model performance and translate theoretical research into production-ready systems.

Responsibilities

Develop novel diffusion models and generative learning algorithms
Design efficient training methods for large-scale generative models
Prototype and evaluate new generative modeling approaches
Collaborate on translating research into production systems
Design rigorous experiments and benchmarks for model quality

Skills

Qualifications

PhD in Machine LearningComputer ScienceStatisticsApplied MathematicsNeurIPS Publication

About Granica

granica.ai

Headquarters

Mountain View, California, United States

Company Size

11-50 employees

Founded

2023

View Profile

Granica is an AI data infrastructure company offering Granica Crunch, a self-optimizing, lossless compression platform that reduces petabyte-scale data storage and cloud query costs by 40–80% across data lakes and engines like Iceberg, Delta, Snowflake, Databricks, and BigQuery. Beyond compression, Granica is building a self-optimizing data factory for AI that includes advanced data sampling, scrubbing, and synthetic data generation to reduce LLM token usage and improve model performance. The company combines production-grade infrastructure with fundamental research, with publications at ICML, ICLR, and NeurIPS.