Data Days Contests | Announcing SQL + AI Promptathon Winners

[slindsay]

Announcing the SQL + AI contest winners!

The SQL + AI Promptathon has officially come to a close, and we're excited to celebrate the incredible work submitted by participants from around the world.

Over the course of the challenge, community members explored how SQL Server, SQL MCP, semantic search, vector embeddings, and AI-powered workflows can be used to solve real-world data problems. From uncovering hidden business risks to testing the limits of semantic search and building entirely new AI-driven solutions, the creativity and technical depth on display were truly impressive.

First and foremost, thank you to everyone who participated. Whether you submitted an entry, experimented with the missions, shared feedback, or followed along throughout the event, you helped make this Promptathon a success. We hope you not only learned something new but also discovered new ways to combine data, SQL, and AI to solve meaningful problems.

You can explore all submissions in the repository here:

👉 https://aka.ms/dd/promptathon

And if you're just discovering the challenge, you can learn more about the event and the missions on the contest landing page: Data Days | SQL + AI Promptathon

🏆 Overall Winner

Wasimrajaanwar, Issue #10

Wasim took on the Data Analyst mission, which challenged participants to uncover Zava's hidden product-quality crisis using sales, support, satisfaction, and customer-feedback data. His investigation moved from category-level risk to SKU-level evidence and uncovered recurring smart-fabric connectivity failures after washing, reported across multiple languages. His thorough analysis, clear reasoning, and compelling use of the available data earned him the highest overall score across all submissions. Congratulations Wasim!

🏅 Mission Winners

Data Engineer Mission

DanWangere, Issue #24

Dan built the required Support Intelligence Mart and then pushed the solution even further. He created and registered a custom semantic-search MCP tool in Data API Builder, generated embeddings directly within SQL Server, and surfaced multilingual defect reports that traditional keyword searches failed to identify.

Data Analyst Mission

Mithun2491, Issue #9

Mithun's analysis identified Premium Apparel as Zava's highest-risk category. By connecting sales data, support incidents, customer satisfaction metrics, and recurring complaints, he highlighted persistent issues related to stitching quality, durability, sizing, fabric performance, and smart-feature reliability.

Data Scientist Mission

tourajshah, Issue #32

Touraj focused on evaluating the trustworthiness of semantic-search results across multilingual customer feedback. His work revealed challenges such as language isolation, misleading product-name matches, and false positives from boilerplate text, offering valuable insights into both the strengths and limitations of semantic search.

Open Mission

rajaechouiref, Issue #11

The Open Mission encouraged participants to define their own challenge and bring their own data. Rajae analyzed 4,899 Capterra reviews and developed a "Liar Score" designed to flag situations where positive star ratings masked negative sentiment within review text. The project expanded into semantic search, recommendation capabilities, and a review red-flag detection system.

Thank You

One of the most rewarding parts of this Promptathon was seeing the diversity of approaches participants brought to the same set of technologies. Some focused on engineering and architecture, others on analytics and investigation, and others on challenging assumptions and validating AI-generated results.

That spirit of curiosity, experimentation, and learning is exactly what Data Days is all about.

Congratulations again to all of our winners, and thank you to everyone who participated. We hope the challenge gave you a chance to get hands-on with SQL + AI, explore new techniques, and connect with fellow members of the community.

We're excited to see what you build next. 🚀