This is a dump of thoughts that's accumulated over the last few months.
I did my first few interviews as an interviewer over the past 2 weeks. I’ve experienced both sides about a dozen times over the past two months. It really comes down to a personality fit as much as a cultural one. A high bar for raw talent makes for a great story, but not always a great organizational move.
Ever since I (re-)learned gradient descent and optimization, I’ve fallen back to it as a good framework for mapping your life trajectory. At any given point, are you moving toward a local optimum, or are you keeping your eyes open for the global one? If you’re pursuing a global optimum, is your step size too large (i.e. you overshoot your target)? Would you settle for a local optimum?
How do we avoid over-training our mental models in real life? It seems impossibly hard, and deep issues stem from something so simple. Racism, stereotypes, miscommunication and misunderstandings can all be viewed as a problem of overfitting. It helps to have some type of awareness and intelligent peers to make sure you’re not over-fittinge.
Practice for me is about training with new examples that grow to cover the domain of inputs. Sometimes the hardest thing is deciding on the training data; the machinery is always there.
I’ve started enjoying life again after grad school. I’m open and hoping for to people and habits to come back. I’m reading slower and longer, writing short and often, and exploring.
I’m still measuring myself, but not constantly. Monthly is a good pace. I need to be humble when making goals.
I love making room for “growth days”. Days during which you make progress and feel stretched on a personal goal.