Engineering Manager at Snappy. 10+ years building software; relocated to the U.S. to build the first U.S. team. Creator and maintainer of Interlace: 77 open-source packages, 1.25M+ npm downloads.
Location
Miami, Florida, USA
Education
CS
Work
Engineering Manager at Snappy · Creator and maintainer of Interlace
The anxiety you describe — "what if the thing I report as a bug isn't actually a bug?" — maps almost exactly to the false positive problem in static analysis. A lint rule that fires too loosely trains developers to ignore it; a competition submission dismissed as "intended behavior" burns the same kind of credibility. In both cases the discipline is the same: can you reproduce it, can you explain the root cause, and can you make the reader believe it matters. The instinct to start with forms and validation isn't naive either — that surface still accounts for a disproportionate share of real exploits, precisely because it's where "I build websites" intuition and adversarial thinking overlap.
I'm a frontend developer who loves turning curious ideas into products and sharing the journey behind them. I write about React, AI, product thinking, and everything I learn while building.
That's the classic rookie mistake. When I used to code in C++ this was my biggest challenge, if I missed one ; then I'm dead. I'll circulate the code multiple times and at last I would found a small semicolon missing. This is was too frustrating as for us we think we wrote the perfect code but reality is we didn't 😆.
Great testing content. One insight from large codebases: the test pyramid still matters — investing in a solid foundation of unit tests with fast feedback loops makes integration and E2E tests far more reliable and maintainable.
Appreciate the testing write-up. For teams starting out, I'd recommend beginning with integration tests rather than unit tests — they provide more confidence with less mocking overhead, and you can add unit tests for complex logic later.
Solid testing patterns. One thing I'd add is the value of proper test data factories over fixtures — they make tests more readable and maintainable, especially when the data model evolves over time.
I'm a frontend developer who loves turning curious ideas into products and sharing the journey behind them. I write about React, AI, product thinking, and everything I learn while building.
Glad the test data approach aligns with what you are building. Once you start using factories to generate edge-case payloads dynamically, you will likely see a massive drop in those annoying flaky tests. Feel free to reach out if you ever want to bounce fixture setup ideas around for your specific stack.
Software engineer building AI developer tools at TokenBay. Passionate about AI infrastructure, open source, and creating great developer experiences. Dog mom to a fluffy Samoyed, League player...
I'm a frontend developer who loves turning curious ideas into products and sharing the journey behind them. I write about React, AI, product thinking, and everything I learn while building.
Well, it was hit and trial. I tried different options and use AI to generate the prompt and use it then. The current cover image is generated using ChatGPT as the Gemini one was more cartoonish. Have you tried generating cover images?
I am a freelance Data Scientist with a proven record of delivering high‑impact, data‑driven solutions that strengthen strategy, security, and innovation.
I'm a frontend developer who loves turning curious ideas into products and sharing the journey behind them. I write about React, AI, product thinking, and everything I learn while building.
I am a freelance Data Scientist with a proven record of delivering high‑impact, data‑driven solutions that strengthen strategy, security, and innovation.
Great article. I've always wanted to try bug hunting but thought I don't have the skills to do so just yet. Looks like I'll drive into it and see where it takes me. Thank you.
We're a place where coders share, stay up-to-date and grow their careers.
The anxiety you describe — "what if the thing I report as a bug isn't actually a bug?" — maps almost exactly to the false positive problem in static analysis. A lint rule that fires too loosely trains developers to ignore it; a competition submission dismissed as "intended behavior" burns the same kind of credibility. In both cases the discipline is the same: can you reproduce it, can you explain the root cause, and can you make the reader believe it matters. The instinct to start with forms and validation isn't naive either — that surface still accounts for a disproportionate share of real exploits, precisely because it's where "I build websites" intuition and adversarial thinking overlap.
I still remember the first bug that took me hours to solve and ended up being a single misplaced character. Humbling experience. 😅
That's the classic rookie mistake. When I used to code in C++ this was my biggest challenge, if I missed one ; then I'm dead. I'll circulate the code multiple times and at last I would found a small semicolon missing. This is was too frustrating as for us we think we wrote the perfect code but reality is we didn't 😆.
Great testing content. One insight from large codebases: the test pyramid still matters — investing in a solid foundation of unit tests with fast feedback loops makes integration and E2E tests far more reliable and maintainable.
Appreciate the testing write-up. For teams starting out, I'd recommend beginning with integration tests rather than unit tests — they provide more confidence with less mocking overhead, and you can add unit tests for complex logic later.
Solid testing patterns. One thing I'd add is the value of proper test data factories over fixtures — they make tests more readable and maintainable, especially when the data model evolves over time.
Sure, thank you so much. I'll keep this in mind.
Glad the test data approach aligns with what you are building. Once you start using factories to generate edge-case payloads dynamically, you will likely see a massive drop in those annoying flaky tests. Feel free to reach out if you ever want to bounce fixture setup ideas around for your specific stack.
I like your cover image lol. It is so creative and I'd like to know how you generated it.
Well, it was hit and trial. I tried different options and use AI to generate the prompt and use it then. The current cover image is generated using ChatGPT as the Gemini one was more cartoonish. Have you tried generating cover images?
Great article, Konark! I love Arnold and the minion. Where is my peter and my luis :)?
You might look like Arnold too.
Here's you fav couple:
hahaha. Is your favourite couple also?
Great article. I've always wanted to try bug hunting but thought I don't have the skills to do so just yet. Looks like I'll drive into it and see where it takes me. Thank you.