Good data
makes better
questions possible.
I’m Shubham, a data engineer building the foundations that help people and AI understand, trust and use data.
From data pipelines
to useful intelligence.
My work sits where data engineering, governance and applied AI meet. These are selected areas of professional work and independent exploration.
Professional work is described at a high level. Independent demos use fictional data and are not replicas of an employer’s systems.
01 / FEATURED FIELD NOTE
Where did this
number come from?
Follow a fictional field from source to analytics, then explore how a description moves through quality checks and human review.
Open Context Lineage Lab ↗INTERACTIVE DEMONSTRATION · SIMULATED WORKFLOW-- An analyst asks a question
$ trace net_claim_value
Engineering the
work behind the answer.
A selection of the problems I’ve worked on at General Mills, shared without internal data, proprietary implementation details or unapproved metrics.
01APPLIED AI / GOVERNANCEMaking enterprise data easier to understand
Making enterprise data easier to understand
The challenge A table name or column label rarely tells the full story. Analysts and AI tools need the context behind the data.
My contribution Led an applied AI initiative for metadata, bringing together data context, generated descriptions and human review. The work progressed into production and is used by multiple teams.
02DATA PLATFORM / MODERNIZATIONBuilding dependable data pipelines
Building dependable data pipelines
The challenge Teams depend on data pipelines that need to be understandable, maintainable and reliable as requirements change.
My contribution Modernized ETL workflows and supported cloud-based orchestration, monitoring and production issue investigation.
03ANALYTICS ENGINEERING / QUALITYTurning raw data into usable models
Turning raw data into usable models
The challenge Analytics and AI need consistent definitions, tested transformations and models that people can understand.
My contribution Developed and supported dbt models, dimensional transformations, documentation and automated data-quality tests.
04APPLIED AI / SELF-SERVICEHelping people ask better data questions
Helping people ask better data questions
The challenge Analysts need accessible ways to discover data and find answers without losing context.
My contribution Built a Glean-powered conversational analytics agent that helps users ask natural-language questions about enterprise data.
05APPLIED AI / PROOF OF CONCEPTExploring unstructured documents
Exploring unstructured documents
The challenge Useful information can be locked inside documents instead of structured tables.
My contribution Developed a proof of concept exploring AI-assisted extraction and analysis of information from documents, with structured output for downstream use.
Selected experience, not an exhaustive project inventory. Descriptions are intentionally high level.
Build the context.
Earn the trust.
Useful AI starts long before a prompt. I think about the data journey, the people who rely on it, and the checks that make its use accountable.
Start at the source.
Understand where data comes from and how it changes along the way.
Make meaning usable.
Pair reliable pipelines with clear models, documentation and business context.
Keep people in control.
Give reviewers the evidence and feedback loops to challenge an output.
Curious by default.
Engineer by practice.
I’m a Mumbai-based data engineer at General Mills, working across cloud data platforms and applied AI. I enjoy the space between a messy question and a reliable answer.
My day-to-day toolkit includes Python, SQL, Google Cloud, BigQuery, Airflow and dbt. I’m interested in data engineering, analytics engineering and AI-ready data foundations.
Have a question
worth tracing?
I’m open to conversations about data engineering and applied AI opportunities.