Shwetank Sheel

I co-founded Accelerate Data to help every data team on the planet ship more.

Shwetank Sheel

What I'm building

We are building Vibedata as a coding agent specialized for data engineering, with a harness for isolation, data-platform guardrails, data-engineering context, and cross-platform reach. See Vibedata.

Experience

Accelerate Data · Co-Founder

Two decades doing this job by hand. Now rebuilding it for agents. That's Vibedata.

Jan 2026 - Present

Rackspace Technology · Director, Global Data & AI Pre-Sales & Solutions

Grew the global data practice's bookings 5x in four years.

Jan 2022 - Dec 2025

Just Analytics · Co-Founder

Bootstrapped it, won six Microsoft and two Databricks ASEAN Partner of the Year awards, and sold it to Rackspace.

Jun 2011 - Jan 2022

Oracle · Senior BI Consultant

Started on Oracle BI and Hyperion, back when data was a nerdy hobby, not a board agenda item.

Jun 2007 - Jun 2011

Recent thinking

A Recipe has already done the explaining

The slow part of working with an agent is explaining what you want. A Vibedata Recipe carries the task, the acceptance contract and the guardrails, and a two-line pointer puts it into an Intent that already holds the context.

16 Nov 2026

Palantir solves the ontology, and the layer beneath it is still your work

Palantir's ontology sits on datasets. The work of getting data in, transforming it and keeping it canonical stays in the layer below.

13 Nov 2026

Ask Vibedata on the pull request and get the commit back

A pull request is already a scoped delegation. On a repository set up for mentions, a reviewer comment reaches the Intent that owns the branch, and the commit comes back to the same review; with no visible Intent, the work arrives as a new pull request.

12 Nov 2026

Your semantic model now answers in your platform's SQL

The semantic query compiles a question in the Domain platform's dialect, including level, time-series and dimension-rule metrics, on DuckDB, MotherDuck, Microsoft Fabric, Databricks and Snowflake.

11 Nov 2026

A report is not a data product

A report is one consumer's answer, built again for the next. A data product is the reusable definition every team and agent reads.

6 Nov 2026

The ephemeral workspace retires the dev/prod pipeline copy

When the target is a runtime coordinate on Fabric, one environment-free orchestration definition can serve an isolated Intent and production. The environments stay. The copies go.

5 Nov 2026

Bound the agent with the semantic model before it touches the lakehouse

On the same column-mapping task, the unbounded run took 38 minutes, the map-first run took 21, and ranking candidates from the semantic model took 5.3. The engineer who ran it attributed the gain to bounding the agent's search.

4 Nov 2026

The promise for each relation now lives in the repository

A data contract is now a canonical artifact in the Domain repository, one list holds personal and install-provided connections, and teams in the Design partner open band can run Databricks notebooks and jobs from the agent.

4 Nov 2026

One advisor's warning: the CDO is liable for the number as well as the context

A banking-industry advisor told us the chief data officer is now exposed for the authenticity of reported data, as well as the context agents receive. Our verifying skill traces a requirement to the assets and tests behind it, then records evidence gathered independently at a named commit, so a team can show why a number is right and say where it could not verify.

2 Nov 2026

Re-engineer a legacy data package one Intent at a time

Your data engineering agent can re-engineer a legacy package one object at a time. Split at the objects that can be tested on their own, give each its own Intent, and its requirement, dbt model and agreed tests are complete before the build.

30 Oct 2026