A Perfect Match: How Brilliant Earth Scaled Data Science with Posit and Snowflake

Summary

As Brilliant Earth continues its rapid growth as a leading digitally native luxury jeweler, its data science and decision intelligence team needed infrastructure that could keep up. With Snowflake as the beating heart of their data strategy, Posit's Native App on Snowflake was a natural fit - replacing one-off processes with a unified, governed platform and bringing AI-assisted development to analysts across the organization.

About:

Founded in 2005 with a mission to make the jewelry industry more transparent and inclusive, Brilliant Earth has spent over two decades proving that quality and conscience don’t have to be in conflict. Today, the company operates a growing number of high-end showrooms all over the world.

Industry:

Retail

Technology used:

Posit Workbench, Posit Connect

Data Cloud Partner Integration:

Snowflake

The challenge

Data Science in Silos, Limiting Enterprise Scale

Brilliant Earth's analytics capabilities had grown organically over time, with the team's data scientists working in silos, utilizing local development environments, limited version control, and hard-to-reproduce workflows. Analysis was spread across traditional BI tools, bespoke data science solutions, and disparate code workflows, making it difficult to scale data science output.

Before Posit, analysts and data scientists were constantly context-switching: jumping between a BI platform for dashboards and a mix of local and online IDEs for ad-hoc analysis, with no clear path to production deployment. No single environment supported both data science workflows and team collaboration, and differing language preferences made code-level coordination hard. Data scientists spent too many hours of their day maintaining and running their data product's code.

The team needed a more collaborative way of working, one that could support professional-grade data science while keeping everything inside a secure, governed environment.

The solution

Posit Team, Natively on Snowflake

Brilliant Earth deployed Posit Workbench and Posit Connect via the Snowflake Native App on Snowpark Container Services (SPCS), giving analysts and data scientists across the organization a single, secure environment with consistent tools and shared governance, all within the Snowflake security perimeter.

In practice, this setup collapses what used to be a multi-step, multi-tool process into a single, continuous workflow. A data scientist logs into Posit Workbench through Snowflake Single Sign-On (SSO), spins up a Positron or VS Code session, and is immediately connected to governed data and the team's code repository. There is no juggling of credentials, no VPN, no separate access requests. From there, the work happens in R or Python with AI-assisted tools, and when a data product is ready to be shared, the team uses Posit Connect for deployment as a scheduled report, an interactive app, or an API endpoint, all in a single click, without leaving the Snowflake security perimeter.

Sharing data products is effortless, as Posit Connect infers access controls from Snowflake, therefore adhering to the company SSO provider. A data science app can be shared with technical and nontechnical users in minutes.

Before, that same workflow might have involved exporting data to a local machine, developing in a disconnected IDE, and then handing off to engineering for production deployment. Now the path from exploration to production is self-serve, governed, and fast.

Realtime Data Science at Scale

Brilliant Earth operates in a high-velocity, trend-driven retail environment where the speed of the marketing engine sets the pace for everything else. The recent Brilliant Earth Ring Pop collaboration is a great example of turning a cultural moment into a data-backed product launch.

When a campaign like this moves from concept to launch, the analytics team needs to keep pace. Previously, evaluating performance meant pulling data from multiple disconnected sources and manually stitching together a picture. Now, the team works from a more integrated, persistent analytics layer where campaign insights are consistent, timely, and ready to act on.

Democratized Access to Analytics

One example of a previously fragmented data product is Brilliant Earth's in-house Marketing Mix Model (MMM). Utilizing Posit Connect, the data science team now surfaces their Marketing Mix Model as a front-end application via Streamlit. This lets Brilliant Earth's Growth Marketing team interact with channel-level performance and scenario planning directly. The underlying data science work happens below the surface, and model updates are pushed live without interrupting the user experience.

The Data Science team has also begun converting AI-generated outputs into Flask API data products deployed on Posit Connect, turning outputs from tools like Claude into persistent, shareable applications the broader team can access on demand.

"The shift isn't just about any single campaign," Data Science Director, Mueller says. "The infrastructure now lets us move at the speed the business requires, while keeping insights governed and reproducible. We can take an AI prototype to a production data product in hours, not weeks."

“Posit products have given my data science and analytics team a single unified foundation, deployed seamlessly and securely via Posit Workbench in Snowflake. On Workbench, we now leverage Positron - which, as a fork of VS Code, felt immediately familiar to our team. The purpose-built features make all the difference: bilingual support for R and Python, an integrated console, integration of Posit’s AI advancements, and a number of other features that result in an IDE designed for data science from the ground up, not retrofitted.”

Matthias Mueller

Director, Data Science

The results

Speed, Scale, and AI That Works With You

Brilliant Earth gained the ability to scale their internal data science products to hundreds of users through automated reporting on Posit Connect. Manual weekly reports were replaced with automated, AI-driven reporting processes. Data science and modeling processes that once lived in manual Python files became automated, interactive financial forecasting apps with Streamlit and Shiny.

The team also built an AI-powered summarization capability using LLMs leveraging the Anthropic and Cortex APIs, generating automated business insights, including campaign ROI summaries for executive leadership. While Mondays were previously spent almost entirely on pulling together prior-week reporting, the analytics team is now able to use that time to focus on impactful deep-dives. Processes that previously required the BI or analytics team to run have become accessible to non-technical users, who can now generate and interact with their own reports and analyses.

The AI capabilities proved to be a key differentiator, not just for the data science team but across executive leadership and nontechnical analysts alike. "The AI capabilities match our expectation as data scientists," Mueller says. "Bringing your own frontier models, Positron Assistant and Databot augment my team's expertise without creating black boxes. It's AI that speeds up how you work, while you stay in control of the output."

Looking ahead

With The Foundation in Place, The Goal is Scale

As Brilliant Earth drives AI adoption further, the team is expanding self-serve analytics so that more stakeholders across the business can access governed insights without waiting in a queue. On the data science side, that means deepening investment in agentic AI workflows, connecting large language models directly to their Snowflake environment so teams can ask analytical questions conversationally and get answers grounded in real data, with full governance intact.

"The vision is a Decision Intelligence layer where Snowflake serves as the analytical foundation, Posit provides the development and deployment surface, and agentic AI acts as the connective tissue between questions and answers," Mueller says.

With Positron, Snowflake Cortex, and Posit Connect embedded in their workflow, Brilliant Earth's data science team can now govern data rigorously, deploy insights rapidly, and bring AI into the process in a way that keeps human experts firmly in control.