What is Snowflake? Developer Demo (2026)

Gilberto Hernandez: Every day, billions of queries are run across petabytes of data to power AI applications, train ML models, and deliver real-time insights to millions of users. Behind it all is Snowflake, a fully managed multi-cloud platform for any data workload. From pipelines and analytics to AI and ML, Snowflake gives builders the tools they need to build data-intensive applications at scale without the infrastructure overhead. That means faster cycles, less grunt work, and more business impact. Let me show you why so many developers trust Snowflake for their mission-critical data workloads.

Vino Duraisamy: If you're a data engineer, you know that building reliable scalable pipelines is the first step to getting value out of your data. Snowflake makes it easy by seamlessly integrating the three critical stages of every data pipeline, ingestion, transformation, and delivery. First, get your data in. Snowflake Openflow lets you ingest any type of data, structured or unstructured, in batches or real-time streams, and from any source. With your data in, it's time to transform it. Whether you're running transformation in Python or dbt projects and SQL queries, Snowflake has you covered. You can build your transformation workflows using user-defined functions, stored procedures, dynamic tables, and more, all in your preferred language. With Python, you can use familiar DataFrame APIs with Snowpark, Pandas, and PySpark to run complex data transformations. And all of your data processing happens inside Snowflake's powerful compute engine. Your code runs next to where your data already lives, which means less data movement and no separate compute clusters to manage. It's faster, more secure, and incredibly efficient. Finally, you need to deliver reliable, high-quality insights. Delivery in Snowflake goes beyond tables and views. You can build fully interactive data applications for your consumers with Streamlit or go a step further and distribute data sets to other developers using Snowflake Marketplace.

Danmei Xu: all developed and shared securely from your own Snowflake account. You can go further with your insights by building agents and complex agentic workflows in Snowflake Intelligence. These agents empower business users to get answers from structured, semi-structured, and unstructured data just by using natural language. Snowflake Intelligence helps you identify and analyze trends, generate visualizations, and get instant answers from your data. If you need powerful, fully managed AI services, then Snowflake Cortex is the foundation you can build on. For example, Cortex Analyst lets you query your structured data with SQL using natural language. It supports semantic views, meaning you can create custom mappings between your business concepts and the tables and columns inside your Snowflake environment. This enables your team to do accurate data exploration at scale. Cortex Search provides a low latency hybrid search engine for unstructured data. It's perfect for building powerful RAG applications and agentic workflows, helping you create an intelligent search experience on top of your data. What if you want to use generative AI directly in your queries, data pipelines, or applications? Using AI-powered SQL functions, you can perform tasks like text summarization, translation, and even sentiment analysis on top of multimodal data sources. You can even invoke your preferred LLM with a custom prompt using your data as context. These functions transform complex AI workflows into simple, powerful SQL statements, unlocking the power of AI for data engineers and data analysts alike. This unified approach to the AI means less time on infrastructure and more time spent on building and scaling your powerful AI applications, all in a single secure platform.

Jacob Prall: Developing your next major application? Snowflake provides an ecosystem of tools and the infrastructure to build scalable, production-grade apps, all while ensuring optimal performance at every stage. For data exploration apps or customer-facing dashboards, Streamlit in Snowflake makes it easy to deploy data apps directly in your Snowflake environment using Python and SQL. Snowflake native apps allow you to build and deploy full featured applications to the Snowflake marketplace, where other developers within Snowflake's ecosystem can install and use them directly in their own Snowflake environment. And for transactional workloads, there's Snowflake Postgres. Snowflake Postgres gives you the full power of community Postgres without the overhead of managing it yourself. It comes ready out of the box with handy extensions to do things like semantic search with pgvector or geospatial queries with PostGIS. Snowflake Postgres also has features like automatic backups and failover with high availability, so you can scale your largest transactional applications with confidence. The rich ecosystem of Snowflake drivers and APIs gives you programmatic access to your data, as well as to AI like Snowflake Cortex. Taken together, Snowflake provides everything you need to launch your high-performance, data-intensive applications.

James Cha-Earley: As an ML engineer, your job is to build and deploy models. Snowflake ML Ops Platform helps you do it faster with full end-to-end support for Python libraries and APIs. You can train and deploy models directly in your Snowflake environment using your favorite libraries like scikit-learn and XGBoost and even bring in your own frameworks. Feature engineering happens right where your data lives in both SQL and Python. For resource-intensive jobs, ML jobs allows you to run workflows on distributed compute nodes with fully managed container runtime. You can work with your preferred IDE like VS Code or Jupyter Notebooks and dispatch your code to Snowflake. This lets you leverage powerful infrastructure, including GPUs, without setup headaches. Snowflake's platform streamlines the MLOps lifecycle from feature stores and managing model versions with Snowflake Model Registry to batch and real-time inference. Your models can consume the same clean, governed data as your analytics teams. Even better, you can deploy models as user-defined functions. This means any team member can get a prediction with a simple SQL query, truly democratizing access to ML within your organization.

Elsa Mayer: At Snowflake, we're deeply committed to open source. Our engineers contribute to projects, collaborate with communities, and incorporate open source and open standards into our platform. We're championing lake house architectures through deep integrations with Apache Iceberg and Apache Polaris. This is about freedom and interoperability. You can write to and query iceberg tables within Snowflake and maintain compatibility across different compute engines without rewriting data or losing historical information. It's truly vendor-agnostic data access. Managing your cloud data platform should be as simple as managing any modern application. That's why infrastructure as code is fully supported through our comprehensive Terraform provider, letting teams manage accounts, databases, compute resources, and security policies, all with familiar DevOps workflows. And of course, Streamlit itself is open source, meaning the apps you build can leverage the entire Streamlit ecosystem, from community components to custom visualizations, while running with Snowflake's scale and security. Our open source commitment doesn't stop there. Open source is embedded across the entire Snowflake platform, with scalable data engineering using Modin, robust ingestion based on Apache NiFi, familiar app foundations with Postgres, efficient machine learning workflows with PyTorch and Ray, and more effective LLMs with TruLens. This open approach allows you to adopt Snowflake incrementally, integrate with existing tools, and maintain the flexibility you require to evolve your architecture without vendor lock-in or data migration overhead, all while benefiting from the continued innovation of open source.

Dash Desai: Powering this entire data experience is Snowflake's unique platform architecture that scales as you build. To start, data storage is completely separate from Snowflake's compute resources, which means you can scale each independently, based on your needs. Storage is automatically compressed and optimized for you. Data in tables is automatically divided into micropartitions, compressed using advanced algorithms and organized for optimal query performance. Compute resources range in size from extra small to 6XL, giving you access to powerful compute as you need it. When not in use, your compute resources auto suspend, so you only pay for what you use. Snowflake runs natively on all three major cloud providers. This means you have flexibility when developing your cloud strategy. Your data pipelines, ML models, AI apps, and more all work the same, regardless of which cloud provider you choose. Security is built into every layer of the platform. Access to your data objects is controlled through fine-grained role-based permissions, giving you complete visibility into who accessed what data and when. Row access policies, masking policies, object tagging, and other industry-leading features help you ensure the highest levels of data governance and compliance for your Snowflake environment.

Biraj Silwal: When you are building with data, you can spend as much time hunting down context as you do actually building. You are tracking down schemas, chasing teammates and digging through docs. And before you know it, half your day is gone and you haven't shaped anything. That's where Cortex Code comes in. It's Snowflake's native coding agent. You tell it what you need in plain language and it works alongside you to get it done. whether that's building a pipeline, exploring your data, training a model, or deploying an agent. It knows your Snowflake environment, so every answer is grounded in your data, your rules, and your business context. It works with the tools you already use, like SQL, Python, dbt, Git, and Streamlit. And because it lives inside Snowflake, your data stays secure. There are two ways to use Cortex Code. You can quickly access it in Snowsight, where you can ask it to write or fix your SQL, explore your data, and manage your Snowflake environment. For example, let's ask it to find the top 10 most expensive queries in the last two days with warehouse and user, and Cortex Code gets to work. or use the Cortex Code CLI in your terminal or favorite IDE. This connects Cortex Code to your local files, code repos, Git, and DBT so you can take on complex end-to-end projects at any scale. For example, let's ask it to build a DBT model for monthly revenue by customer segment. It works through each step to scaffold the project aware of your schema and business logic from the start. You can debug and automate pipelines end-to-end, train and deploy models, build and optimize AI agents. The possibilities are endless. And with customizable skills, you can even extend Cortex Code to work exactly the way your team needs. Anything you want to do with data, do it confidently with Cortex Code.

Gilberto Hernandez: So here's what this means for you as a builder. Rather than spend time on infrastructure, data movement, and general setup challenges, you can focus on building and creating value all within a unified, fully managed data platform that supports your data workloads and tons of complex use cases. So whether you're building data pipelines, training ML models, creating intelligent applications, working in open data ecosystem, or analyzing massive data sets, Snowflake provides a single platform that scales with whatever it is you're building. If you're ready to build something amazing, start with a free trial account at signup.snowflake.com and check it out for yourself. What are you planning to build?