Letโs build a benchmark to benchmark all AI benchmarks. I will say the criterias should be reproducibility, coverage, fairness, and robustness. DM, lol!
Absolutely not, can be dead for your own use-case maybe?
1. Cost effective
2. Focused context
3. Lower hallucinations
4. Realtime updated knowledge base.
Interviews need to be fixed! modern software engineering isnโt about remembering syntax or doing everything from scratch. Itโs about knowing how to solve problems effectively with the best tools available.
Here are a few things I like to do:
1. Identify the bottlenecks
2. Take external openings, from people with more experience than me.
3. Run experiments, before finalising what needs to be done.