My Journey in AI: From Beginner to Builder
Hi everyone! Welcome to my developer and founder blog, where I share my ongoing projects and share a little about how I got here. It started back in November 2022, when I first started playing with GPT-3. I took my first steps into this rapidly evolving field. Fast forward to today, and I've gained valuable experience, built some interesting projects, and even contributed to the ChatGPT ecosystem.
My journey actually begins with my day job. I work as a building automations controls specialist in commercial buildings. I am responsible for managing the logic and operation of the lighting and HVAC. While this doesn't involve any coding language knowledge any more, I still gained valuable experience. So technically for last decade, I have been playing with programming logic and incredibly complex work flows. Naturally my next step was to jump into development with no-code/low-code platforms like Bubble.
Accidental Innovation:
My first project was something I called "Fabric." It consisted of some wrappers around the GPT API, accomplishing a variety of text generating tools for small businesses. Fabric included writing copy for websites, ads, etc. I developed some more specialized tools however:
- Sales Script: Allowed users to upload a transcription of a conversation and have a script of responses and some analytical insights into the conversation to prep for a followup.
- LitNav: I pulled in the Google Books API and used book content and cover images to write summaries of any part or of the book as a whole. These summaries included key terms, concepts, actionable next steps, and some external related quotes and links. It was meant for busy business owners to digest books as they worked. What I didnt realize is that I had created one of the first RAG applications for AI, MONTHS ahead of anyone else.
- Product Engine: This was my crown jewel. Product Engine allows users to enter in a product name, ad copy, description, key words...any one part of an idea or service, and it would fill in all the rest. Then I had it write an image description and fed that back through Dall-e 2. So in one click it created a product prototype and a mockup image. I had created one of the first MULTI-MODAL applications.
Needless to say I was HOOKED. I realized that all my previous work had taught me exactly how to create back-end logic, and I started building some of the first prompt-chain frameworks in Bubble. I continued to build and learn by talking with ChatGPT. I've done much more since, I'm going to continue this story, so I hope you'll check back!
In the meantime, I'm starting a series on OpenAI's upcoming SORA model. Ill be taking each research paper cited in their work and using PaperSnap.ai to create some digestible summaries and posting SnapCast audio versions of each one. Knowledge is power, and I want to help you level up!
Best Wishes, Warmest Regards
Mark Z.
