skip to content

Sanveer Singh Osahan

I build customer value through engineering, leading teams that ship what customers trust and businesses grow on.

I'm a software engineer, and the Data + AI ecosystem is where I love to work. I build the systems that make data trustworthy at scale: data platforms, cataloging, lineage, and the extraction pipelines underneath them all.

The problems I like only show up at scale. At Atlan, I designed the industry's first incremental metadata extraction, taking extraction pipelines from hours down to minutes. It went on to power Atlan's Snowflake, Databricks, and Oracle connectors for over a hundred customers. Before that, I was a founding engineer at Boltic, building the core of a SaaS ETL platform.

I care about two things: the customer on the far end of the system, and the engineers building it. A lot of my work is mentoring, helping people reason about system design, performance, and the quality bar that makes software worth trusting by the customers using it.

Databricks is the platform I go deepest on. I helped strengthen Atlan's partnership with Databricks, working with their product and engineering teams to ship the System Tables for Lineage and Query History that our shared customers now derive real value from. I follow what they ship closely and enjoy seeing how their newest pieces solve real problems in the Data + AI space.

Articles

Notes

AI Coding Agents are not capable enough to build a delightful experience for end users. They just spit out code and lack that human touch.

When building for humans, the core aspect to earn their trust is care. Care is a human trait and nothing will be able to take that away from us.

The positive side to using the Coding Agents:

A few years ago, when I tried building some products as a side project, being a backend developer, I struggled with frontend. Got overwhelmed with what JavaScript library to learn — React, Angular, Vue, Next — instead of focusing on the build.

Now the Coding Agents can write the frontend for me. And I can focus on what really matters — building a delightful experience for my users.

Whether you are a backend, frontend, full stack, mobile, or an AI engineer, don’t forget to put that human touch of “care” in your work. That will drive you to be better, to grow, not only in your career, but also in your life.

Care is both a deliberate action and an emotion so strong no AI model or harness will be able to capture it.

When building a product, do not wait for others to find basic issues, least of all the customer. Don’t leave them for your team to clean up either. Build it, own it, and validate it.

Here is a simple routine I follow:

Build the “happy path” toward the end of the day. End your day on a high note, confident that you have built a world-class feature, and get a solid 7 - 8 hours of sleep.

The next morning, wake up and step entirely into the shoes of your user. If you encounter an annoying experience, do not worry, your builder persona is there to fix it.

The more effectively you can switch between these personas, the better the final product experience will be. Your user persona must never carry the biases of your builder persona.

There is something appealing about still watching tutorial videos or reading documentation where the steps being performed have an explanation of what we are doing and why we are doing it.

Coding agents may have made it faster to set up and execute, but suddenly you will see that your solution is using a webhook and you have no idea why.

This adds a comprehension debt, and you can’t optimise or scale a system that you do not understand.

A good strategy is to offload but be curious to learn. Get your solution to work first. You also can’t optimise or scale a solution that does not work.

See all notes

From my bookshelf