Research Scientist — Tabular Data Learning

Granica is redefining how enterprises prepare and optimize data at the most fundamental layer of the AI stack—where raw information becomes usable intelligence. Our technology operates deep in the data infrastructure layer, making data efficient, secure, and ready for scale.

We eliminate the hidden inefficiencies in modern data platforms—slashing storage and compute costs, accelerating pipelines, and boosting platform efficiency. The result: 60%+ lower storage costs, up to 60% lower compute spend, 3× faster data processing, and 20% overall efficiency gains.

Why It Matters
Massive data should fuel innovation, not drain budgets. We remove the bottlenecks holding AI and analytics back—making data lighter, faster, and smarter so teams can ship breakthroughs, not babysit storage and compute bills.

Who We Are

• World renowned researchers in compression, information theory, and data systems

• Elite engineers from Google, Pure Storage, Cohesity and top cloud teams

• Enterprise sellers who turn ROI into seven‑figure wins.

Powered by World-Class Investors & Customers
$65M+ raised from NEA, Bain Capital, A* Capital, and operators behind Okta, Eventbrite, Tesla, and Databricks. Our platform already processes hundreds of petabytes for industry leaders

Our Mission: We’re building the default data substrate for AI, and a generational company built to endure.

Smarter Infrastructure for the AI Era

We make data efficient, safe, and ready for scale—think smarter, more foundational infrastructure for the AI era. Our technology integrates directly with modern data stacks like Snowflake, Databricks, and S3-based data lakes, enabling:

• 60%+ reduction in storage costs and up to 60% lower compute spend

• 3x faster data processing

• 20% platform efficiency gains

Trusted by Industry Leaders

Enterprise leaders globally already rely on Granica to cut costs, boost performance, and unlock more value from their existing data platforms.

A Deep Tech Approach to AI

We’re unlocking the layers beneath platforms like Snowflake and Databricks, making them faster, cheaper, and more AI-native. We combine advanced research with practical productization, powered by a dual-track strategy:

• Research: Led by Chief Scientist Andrea Montanari (Stanford Professor), we publish 1–2 top-tier papers per quarter.

• Product: Actively processing 100+ PBs today and targeting Exabyte scale by Q4 2025.

Backed by the Best

We’ve raised $60M+ from NEA, Bain Capital, A Capital, and operators behind Okta, Eventbrite, Tesla, and Databricks.

Our Mission

To convert entropy into intelligence, so every builder—human or AI—can make the impossible real.

We’re building the default data substrate for AI, and a generational company built to endure beyond any single product cycle.

WHAT YOU’LL DO

• Invent and prototype new algorithms that advance representation learning and compression for structured and tabular data at petabyte scale

• Design cost models and adaptive learners that fuse statistical learning theory with systems-level optimization

• Develop novel architectures and approximation schemes that enable efficient inference and training on heterogeneous enterprise data (structured, semi-structured, unstructured)

• Create telemetry-driven encodings and embeddings that continuously adapt to real-world data distributions

• Partner with Montanari’s research group and Granica’s systems engineers to translate new learning methods into production-grade services used by customers like Lyft, Pinterest, and Snap

• Author world-class research papers and design docs, mentor peers, and open-source algorithms where possible

• Contribute to the scientific community by publishing results that bridge theory, tabular learning, and scalable infrastructure

WHAT WE’RE LOOKING FOR

• PhD in Machine Learning, Statistics, or related mathematical field with focus on representation learning, generalization, or structured data modeling

• Publications or applied research in areas such as tabular data learning, feature learning, or multimodal fusion

• Strong foundation in optimization, information theory, or statistical learning

• Hands-on experience with deep learning frameworks (PyTorch, JAX, TensorFlow) and Python / Rust for high-performance prototyping

• Track record of validating ideas against large-scale, real-world datasets or production systems

• Pragmatic researcher who seeks elegant, empirically validated approaches to hard data problems

WHY JOIN GRANICA

This is a rare opportunity to build the foundations of learning systems for structured data, working directly with Stanford Prof. Andrea Montanari, one of the leading figures in modern statistical learning theory.

What’s compelling for engineers with deep algorithmic backgrounds:

• Research that ships—you’ll invent theory and implement it in real-world, PB-scale systems

• Full-stack math—from cost models and combinatorics to vectorized kernels and scheduling heuristics

• Fast feedback loops—prototype overnight, validate on live traces the next day

• Extreme ownership—partner directly with core systems engineers and product leads to bring your algorithms to life

• High-impact culture—every improvement is measurable in dollars saved, seconds shaved, and human time unblocked

COMPENSATION & BENEFITS

• Competitive salary and equity

• Premium healthcare, vision, and dental

• Unlimited PTO + quarterly recharge days

• Stipends for learning, publishing, or open-source engagement

Equal Opportunity

Granica is an equal opportunity employer. We’re committed to building a diverse, inclusive team and encourage candidates from all backgrounds to apply.