Results for “find relationships in interview”

Chase lays out when to fine-tune (behavior/style) vs. when to RAG (knowledge). Fine-tuning teaches how to speak; RAG teaches what to say.

Hightower uses a restaurant kitchen analogy: pods are dishes, nodes are stations, and the scheduler is the head chef assigning work. Deployments handle rolling updates without downtime.

Willison walks through hybrid search: dense embeddings capture semantic similarity while BM25 catches exact terminology. The best production systems combine both with reranking.

Swyx explains that long context windows don't solve the 'needle in a haystack' problem — models still lose focus. Retrieval remains essential even with 1M-token windows.

Naval defines leverage as force multipliers: labor (people), capital (money), and code/media (permissionless, zero marginal cost). The last two scale without your direct time.

Andreessen defines PMF as being in a good market with a product that can satisfy it. Before PMF, everything feels like pushing a boulder uphill. After, it's pulling.

Newport assigns every minute of the workday to a specific task block. If it's not on the calendar, it doesn't happen. Reactive work gets squeezed into explicit slots.

Clear explains the two-minute rule from Atomic Habits: if a task takes less than two minutes, do it now. The goal isn't the task — it's becoming someone who takes action.

Graham distinguishes maker schedules (long uninterrupted blocks) from manager schedules (hourly appointments). A single meeting can destroy an afternoon of creative work.

Huberman clarifies that dopamine drives anticipation and pursuit, not satisfaction. The spike happens before and during the chase — which is why achievement often feels hollow.

Dweck defines fixed mindset as believing abilities are static, and growth mindset as believing abilities develop through effort. Praise for effort, not intelligence, shapes which mindset children adopt.

Graham argues early startups should manually recruit users one by one. Stripe's 'Collison installation' — setting up accounts on the spot — is the archetype.

Livingston distills startup failure to two causes: you run out of money, or founders lose conviction. Most 'product' failures are actually motivation failures in disguise.

Norton argues most adults under-eat protein. 1.6g/kg supports muscle retention during fat loss, and protein has the highest thermic effect of any macronutrient.

Irvine describes negative visualization: periodically imagining loss of health, relationships, or possessions. Counterintuitively, this increases gratitude and reduces anxiety about the future.

Karpathy breaks RAG down as retrieval-augmented generation: instead of stuffing everything into a model's context window, you fetch relevant documents first, then let the LLM synthesize an answer grounded in those sources.

Kahneman discusses how people overestimate how much others notice their appearance and behavior. The 'invisible gorilla' experiments reveal how blind we are to our own inattention.

Huberman explains that viewing bright light within 30-60 minutes of waking sets your circadian clock, boosts cortisol at the right time, and improves sleep quality that night.

Harris explores Korzybski's insight that our models of reality are not reality itself. Confusing the two leads to ideology, tribalism, and the inability to update beliefs with new evidence.

Huberman explains cold exposure triggers norepinephrine and dopamine increases lasting hours post-exposure. The key is deliberate discomfort with controlled breathing — not hypothermia.