Pictured above, you see me in my enlightened repose; once a mere mortal man, now a super-powered cybernetic being. For I have harnessed a great power: vectorizing unstructured text data into a database and using Natural Language Processing to provide me insights.
Ok, pardon the hyperbolics. In all earnestness, the biggest success of this course for me was simply acquainting myself with the field and toolset of AI. Completing my project will give me the opportunity to include this in-demand field on my resume. The field is intensely interesting to me, so I hope that this comes to fruition in my job search. Indeed, at my internship I have already been tasked with researching and improving an AI feature.
I don’t think I can pose this success as a “breakthrough.” It was more like a series of incremental learnings. I suppose this is probably true for most “breakthroughs.” But the point stands: at no point were me and my group stuck on a single concept, and then elated to have overcome it. The process was more like a long series of small breakthroughs. No one was more significant than another. Cumulatively, at the other side of this completed project, one can consider the completion of the whole task as a breakthrough. But that would be misleading. It was more like a gradual clearing of fog. We would learn enough about the process to understand our next step of implementation. Once we moved forward to that next step, we have enough context to see forward to the next step. We could only vaguely wave our hands at any consequent steps after that. We might have an idea of how we were going to implement the step after next, but then we’d gain understanding and context, and realize that it would have to be done differently.
Now that we have this perspective, I’m looking forward to my next project involving these tools. Or, I’m also looking forward to continuing to iterate this project and refine it. I believe in the window of time that we had for this semester, we were able to piece together what amounts to a prototype. It taught us many things. When we initially wondered how to design our vector database schema, we merely wanted to get a working implementation. Now, we could revisit those decisions and understand their consequences. Perhaps we’d store attributes that were once together in a more granular fashion, and this would allow for more diversity in the vectorization of our data.
As a sidenote, a bigger “breakthrough” might have also been a little more mundane. This is the first project where I’ve worked with a team and a docker-compose. Using docker is very common, and often a required skill on job applications. Before I had made passing attempts to incorporate Docker into my workflow. This time it finally stuck, and I used it successfully. Up until this point, I didn’t feel justified in putting this tool on my resume. Now I’ve added it, and I think it will be one more card in my hand at the employment-poker-table.
Leave a Reply