Vicki Boykis
Machine learning engineer working on recommender systems, personalization and information retrieval; author of the 'What are embeddings?' text.
Vicki Boykis did not write this page.
We collected these quotes from things they published elsewhere, and every quote links to where it was said. They have no account here and have not endorsed this site. Quotes are word for word; the short line under each one is our own restatement, not their wording. Their own site. Is this you? Claim it or ask us to remove it. Or tell us what is wrong here.
Filter All topics · Newest first
-
Their wordsAn app feature, once added to a codebase, can be extremely hard to remove.
-
Their wordsDeleting code is hard, particularly because it's a thankless task that usually doesn't go on promo packets, which is why we end up shipping hundreds of thousands of lines of sparkle emoji magic wand AI features, but deleting code is silent.
-
AI slopcode generationsoftware quality
Their wordsBut in this new era, we must delete code even harder, because if we don't constantly remove complexity, we are going to drown in generated kudzu slop that overtakes codebases.
- 3 weeks earlier
-
Their wordsBut where it starts to get dicey is when you have both jargon and long text explanations being generated by machines, for people at different levels of understanding.
-
Their wordsThe output is the result of a model trained on the entire corpus of technical internet documentation with the training objective being completing a sentence but not with the objective of human understanding.
-
Their wordsWe should strive for simplicity. If we can't, it could be because we don't understand the problem ourselves yet enough to compress it.
- 8 weeks earlier
-
Our reading
Local language models can now perform agentic coding tasks at close to frontier-model quality.
Their wordsI've finally been able to do agentic coding locally and have loops work at about ~75% the accuracy/speed of frontier models, which is incredible.
↗Running local models is good nowvickiboykis.com 1st of 3 in this piece
-
LLMsdeveloper toolssoftware quality
Their wordsNeedless to say, I'm not sure this is ready for production software development quite yet.
↗Running local models is good nowvickiboykis.com 2nd of 3 in this piece
-
Their wordsThe benefits, though, are numerous and the ecosystem critical to invest in, particularly now.
↗Running local models is good nowvickiboykis.com 3rd of 3 in this piece
- 3 weeks earlier
-
Our reading
AI code generation's default reward loop works like a slot machine and undermines skill retention.
Their wordsCode generation, in its default mode, is antithetical to skill retention, particularly because its UX affordances are reminiscent of a slot machine's: you pull the lever, you get a reward (a solution to your coding problem.)
↗We should be more tired than the modelvickiboykis.com 1st of 2 in this piece
-
Their wordsIt really does take extra concerted effort to move from just generating answers to using the tool deliberately.
↗We should be more tired than the modelvickiboykis.com 2nd of 2 in this piece
- 17 months earlier
-
Their wordsYet, I also need to read fiction because if I only read tech books, they take away my creative energy rather than grant it, as fiction does.