Personal Blog

What I learned

Short notes from things I built, broke, fixed, and learned.

01

Building an AI chatbot

Building the chatbot was not the hard part. Making it useful and reliable was. I used LangChain and OpenAI APIs, improved the flow, and cut weak responses until it felt like a product people could trust.

  • Good AI UX is mostly product judgment.
  • A 40% lift came from many small fixes.
  • User trust matters more than clever prompts.
02

Adding voice to a mobile app

I added Whisper-based speech-to-text and text-to-speech to a React Native chatbot. It made the app feel more natural, but it also exposed a lot of small edge cases very quickly.

  • Voice can make an app feel faster.
  • Hands-free input improves mobile UX.
  • Audio features come with many edge cases.
03

Simple systems that mattered

Some of the least flashy systems turned out to be the most useful. Working on an OCR inventory tool and an energy dashboard reminded me that people mainly want software that makes daily work easier.

  • Internal tools matter because confusion is expensive.
  • Clear dashboards save real time.
  • Useful beats flashy.
04

Learning to ship smaller

I used to think progress meant building a big feature all at once. Shipping smaller taught me a better rhythm: make one useful slice, test it, listen to the rough edges, then keep going.

  • Small releases make feedback easier to trust.
  • A narrow scope can still feel complete.
  • Momentum is built by finishing real pieces.
05

Designing for the person in a hurry

The best interfaces I worked on were not the loudest ones. They made the next step obvious, kept important details close, and respected the fact that users usually arrive with something else on their mind.

  • Clear hierarchy is a feature.
  • Good defaults reduce nervous clicking.
  • Calm screens make complex work feel manageable.

More posts will come as I keep shipping and learning.

How'd this one land?

Loading exact counts...