Harsha Vundavalli

Developer

[Harsha Vundavalli @harshavundavall]

❄️ Snowflake in one line: your data warehouse, data lake, and AI workloads in one platform, with storage and compute scaled independently. Spin up a warehouse, run your query, pause it. You pay for what you use.

#Snowflake #DataEngineering #CloudData

[Harsha Vundavalli @harshavundavall]

The most useful thing I do with Cortex Code is argue with it.

It wanted synthetic data. I found 2.2B rows on Marketplace. It said no ServiceNow connector existed. It did.

CoCo drafts, I challenge.

#Snowflake #CortexCode

[Harsha Vundavalli @harshavundavall]

Interactive Warehouses target low-latency, high-concurrency workloads, while Adaptive Compute helps optimize compute for changing demands. The benchmarks showing up to 9x higher throughput and 3x lower latency were particularly interesting.

#Snowflake #DataEngineering @Snowflake

[Harsha Vundavalli @harshavundavall]

❄️ Great time at the Snowflake Austin User Group meetup!
Really enjoyed the discussions on Snowflake Native Apps, Agentic Entity Resolution, Snowpark Container Services, Adaptive Compute, and Snowflake Postgres.
Always good connecting with the Snowflake community #snowflake

[Snowflake @Snowflake]

Testing Apache Iceberg™ streaming at 1M+ TPS shouldn’t require building the surrounding infrastructure first.

Snowpark Container Services generates the load, Snowpipe Streaming batches it server-side, and Snowflake lands and compacts the Parquet files.

See the architecture: https://t.co/e0IduLfqGv

[Snowflake @Snowflake]

Snowflake helps @bookingcom teams centralize data in a discoverable catalog, making it easy to find, trust, and deploy across data and AI workflows.

See how Snowflake helps data and AI do more: https://t.co/0UFSAunslT

[Snowflake @Snowflake]

SQL. Python. AI agents. GPU pipelines.

One platform.

Self-optimizing compute that brings resources to your data – automatically, without moving or duplicating it.

Adaptive is now GA on AWS, Azure, and GCP.

[Snowflake @Snowflake]

Most retailers run BI and advanced analytics on two separate systems. @UnderArmour runs both on one. ⚡

Jeff Leineweber, Director of Data Platform Engineering, shares how that let his team move faster and share data across their ecosystem in a fraction of the time. See how 👇 https://t.co/RZRV2J3x9T

[Snowflake @Snowflake]

PostgreSQL 19 is delayed after 53 reverts since beta began.

One example: GROUP BY ALL was pulled after post-commit review found it could produce wrong results.

That isn’t beta failing. It’s beta doing its job. See what was reverted, what remains at risk, and what is still expected to ship: https://t.co/ryrjtK3GpH

[Snowflake @Snowflake]

The best AI doesn't move your data to the model. It runs the model where the data lives.

That's what @_PowerDigital_ and @thoughtspot are doing with Snowflake on @AnthropicAI's Claude Marketplace – governed AI, grounded in business context, inside the platform their teams already trust.

[Snowflake @Snowflake]

More model choice means more flexibility to match the right model to each task.

Kimi K3, Moonshot AI’s open-weight model, is now in private preview on Snowflake Cortex AI for coding, research, and analytical workflows.

Learn more https://t.co/epWSrPQv48

[Snowflake @Snowflake]

Intelligence is only half the job. Acting on it safely is the other half. ⚙️

@Xero's James Hsu joins Snowflake's Anahita Tafvizi, sharing why AI agents need governance like access control and masking built in natively, not bolted on. Watch here 👇 https://t.co/hp8LernkYz

[Snowflake @Snowflake]

.@Get_Writer + Snowflake – a closed loop that compounds with every campaign. With WRITER's MCP connector – powered by Cortex Agents – governed segments, LTV scores, and churn signals go straight from Snowflake into the hands of every marketer, not just the data team. Briefs, assets, copy, campaigns – all generated on-brand, on-policy, and on the data your org already trusts. Customer's like Beautypie are already seeing it transform their marketing campaigns.

[Snowflake @Snowflake]

Streamlit in Snowflake Workspaces separates editing from publishing.

Run changes privately in a development app. When they’re ready, deploy them to the Streamlit object your users access.

Collaborate through Git-backed or shared workspaces: https://t.co/Q1IR5QLyLh

[Snowflake @Snowflake]

Adding a new entity type to a knowledge graph usually means changing the schema.

Snowflake Data Superhero Satish Tirumalasetti shows a metadata-driven approach on Snowflake: KG_NODES + KG_EDGES store the graph, while a stored procedure generates typed views.

See the implementation: https://t.co/ioByQNvyrf

[Snowflake @Snowflake]

Sky’s the limit when you launch your career in the AI Data Cloud. ☁️

Just ask Snowflake’s interns. From leveling up fast with leading enterprise AI tools to impactful work actually defining the AI era, it’s an internship to remember (and one you’d want to do again).

Whether you’re coming back for more or want to experience #LifeAtSnowflake for the first time, spring 2027 internships are now open: https://t.co/PSVr3AQqlX

[Snowflake @Snowflake]

Governance used to mean slower data and AI initiatives. Not at @BookingCom

With our integrated data governance, https://t.co/PQkoNbio1H teams have a trusted foundation to find, access, and use fully compliant data faster. https://t.co/toVQoAFBfX

[Snowflake @Snowflake]

One more model drop, with two new choices now in public preview on Snowflake Cortex AI.
Reach for GPT-6 Sol when a task needs deeper reasoning across complex professional and technical work, and GPT-6 Luna when you need responsive execution at scale.

Available now: https://t.co/33hc6Zntlh

[Snowflake @Snowflake]

The model already knows dbt. A skill should capture what it can’t infer: your naming conventions, partitioning strategy, and exact team workflow.

Encode that knowledge once, then reuse it across CoCo sessions: https://t.co/xMdYGPQ5lM

[Snowflake @Snowflake]

The next competitive advantage won't come from collecting more data, it will come from activating the intelligence you already have. 🚀

See how Saga bridges the gap between secure cloud data and marketing execution using AUDIENCES and Snowflake. By activating first-party audiences directly from their cloud environment, they drove a 10.9x incremental ROAS and a 65% lower cost per lead on recent campaigns.

Read the full story on how to turn data into measurable marketing growth: https://t.co/FMybg8sYuy