Satya malugu
@prolificcoder.bsky.social
Mobile, flutter, SoFi, chess national master, test automation, individual freedom, badminton, science fiction
Misery likes company. I have been paying high taxes for ~20 years without a say in government.
January 8, 2026 at 3:41 PM
Misery likes company. I have been paying high taxes for ~20 years without a say in government.
At SoFi we are experimenting with several AI/llm for developer productivity and three specifically for code review with in MRs. Can expand more privately
January 5, 2026 at 12:00 AM
At SoFi we are experimenting with several AI/llm for developer productivity and three specifically for code review with in MRs. Can expand more privately
Definitely agree. Our mobile repo test volume has increased 10k from 20k just in last 3 months. And coverage has gone up too.
Just have to be careful not to let it slop in.
Just have to be careful not to let it slop in.
December 20, 2025 at 10:48 PM
Definitely agree. Our mobile repo test volume has increased 10k from 20k just in last 3 months. And coverage has gone up too.
Just have to be careful not to let it slop in.
Just have to be careful not to let it slop in.
Also shamelessly ignoring that they are eliminating genuine competition through coercion. Competition gives better results for consumers and the economy.
September 21, 2025 at 10:58 PM
Also shamelessly ignoring that they are eliminating genuine competition through coercion. Competition gives better results for consumers and the economy.
Is he shaking hands with Tal?
August 17, 2025 at 8:50 PM
Is he shaking hands with Tal?
I don't know the reality but I feel like reducing unnecessary digital crap reduces clutter and might help storage overall. Like I am not going to look through all the 50 images of a beach trip 7 years ago. Is better to reduce for both human and digital organization
August 12, 2025 at 11:41 PM
I don't know the reality but I feel like reducing unnecessary digital crap reduces clutter and might help storage overall. Like I am not going to look through all the 50 images of a beach trip 7 years ago. Is better to reduce for both human and digital organization
So by deleting them you are reducing the need for keeping that storage alive. Demand for storage keeps growing linearly, any reduction will help.
August 12, 2025 at 11:14 PM
So by deleting them you are reducing the need for keeping that storage alive. Demand for storage keeps growing linearly, any reduction will help.
Sorry? Why won't it? If say a data center has to store 100 peta bytes today but say based off all these deletions they only store 50pbs
That's saving. Data has backups and redundancies. It might help create extra capacity for future storage.
That's saving. Data has backups and redundancies. It might help create extra capacity for future storage.
August 12, 2025 at 8:51 PM
Sorry? Why won't it? If say a data center has to store 100 peta bytes today but say based off all these deletions they only store 50pbs
That's saving. Data has backups and redundancies. It might help create extra capacity for future storage.
That's saving. Data has backups and redundancies. It might help create extra capacity for future storage.
At SoFi (~1500+) we are an AI experiments team and we are piloting several tools including cursor, Claude, windsurf, gitar etc. we started with copilot, got Gemini first but then but our CTO opened the gates few months ago.
July 16, 2025 at 11:41 AM
At SoFi (~1500+) we are an AI experiments team and we are piloting several tools including cursor, Claude, windsurf, gitar etc. we started with copilot, got Gemini first but then but our CTO opened the gates few months ago.
Context matters but this seems overall bad advice. In large enterprises and code bases there is a lot of tech debt and not everything needs to be solved right away. Especially not by your highest paid engineers. They should go after the hardest and most impactful problems.
July 7, 2025 at 11:03 AM
Context matters but this seems overall bad advice. In large enterprises and code bases there is a lot of tech debt and not everything needs to be solved right away. Especially not by your highest paid engineers. They should go after the hardest and most impactful problems.
So which cache was it? Did you delete derived data folder?
July 7, 2025 at 10:59 AM
So which cache was it? Did you delete derived data folder?
Are your using cursor? I think the most useful way is to create cursor rules to do your steps and publish them in GitHub. You can double check the rules. So future users can just those rules and direct their agents to do the same. A video is for humans but rules are for AI.
July 1, 2025 at 2:13 PM
Are your using cursor? I think the most useful way is to create cursor rules to do your steps and publish them in GitHub. You can double check the rules. So future users can just those rules and direct their agents to do the same. A video is for humans but rules are for AI.
We still have them in git repo, so you think we can remove them
June 22, 2025 at 7:57 PM
We still have them in git repo, so you think we can remove them
Don't you need a pubspec.yaml /.lock in each of the modules too?
June 21, 2025 at 8:34 PM
Don't you need a pubspec.yaml /.lock in each of the modules too?
Why not movie and audio generation?
May 22, 2025 at 5:11 PM
Why not movie and audio generation?
Oh small world I saw this line from my taxi to work and taxi driver is like there is some party, people are lining up! Its a MCP party
May 15, 2025 at 1:29 AM
Oh small world I saw this line from my taxi to work and taxi driver is like there is some party, people are lining up! Its a MCP party
Remember you have to create force update strategy well BEFORE you need it.
May 6, 2025 at 12:51 PM
Remember you have to create force update strategy well BEFORE you need it.
Any E2E automation tool will be slow, dont compare it to grey box tools like espresso
May 2, 2025 at 11:44 AM
Any E2E automation tool will be slow, dont compare it to grey box tools like espresso
We are using them to automate deep link tests. They need access outside the app and maestro is very simple for that usage.
Let's not forget that it's an E2E framework and we shouldn't try to introduce mocking that's better left for lower level tests.
Let's not forget that it's an E2E framework and we shouldn't try to introduce mocking that's better left for lower level tests.
March 21, 2025 at 11:04 AM
We are using them to automate deep link tests. They need access outside the app and maestro is very simple for that usage.
Let's not forget that it's an E2E framework and we shouldn't try to introduce mocking that's better left for lower level tests.
Let's not forget that it's an E2E framework and we shouldn't try to introduce mocking that's better left for lower level tests.