I'm giving a talk at YC @startupschool later this month.
We'll go deep on what it takes to optimize LLMs for latency and throughput.
If you're cracked, or just interested, come say hi! We're scaling @luminal_ai's engineering team rapidly (min 10 hires by the end of the year)
How to rent an NYC apartment without a US co-signer?
Your options: upfront rent, university housing, or a lease guarantor service.
For most renters, the fastest and most reliable path is guaranty insurance.
Get an instant quote → https://t.co/FN2qcKr6mm
Underrated life advice: When you're feeling overwhelmed, there are only two things you should do... get organized and get to work. The rest is just noise. Peace is found in progress.
Real winning comes from knowing the hard work and sacrifice you put in resulted in your success.
Most people want to win without the work part, which is a child’s idea of winning.
@Electrarythm “Modern man finally drags a huge crowd of indigestible rocks of knowledge around inside him…”
— Friedrich Nietzsche’s 1874 essay, "On the Use and Abuse of History for Life"
The legacy rent guarantor companies built their businesses for a world where landlords and renters had no other options.
That world is gone.
PandaGuarantee: Fast same-day tenant approvals. Claims paid in just 3 days. Usually 20% cheaper (or more) for renters.
Legacy guarantors have cumbersome applications, slow processes and expensive pricing.
Better, Faster, Cheaper is here! 🐼
No idea if this will get any reach to help someone but…
things I’m learning as a second-time founder:
1. Don’t try to avoid all mistakes… instead, make faster mistakes. I learned the most growth comes from just trying more things with low expectations and being surprised and curious about what’s working. Stay open to being surprised.
2. Be opportunistic. I’ve had the most success when I simply embrace clear opportunity signals and then fall in love with what a market really wants. Everything becomes infinitely interesting if you stay curious: the most banal-seeming business reveals beauty once you dive into its history, industry nuance, colorful characters, and the real human needs it touches. Genuine fascination compounds into persistence which is the best unfair advantage.
3. Past success can distort your signal.
After you’ve had one win, people are more likely to nod along. That feels nice, but it’s dangerous. Overly deferential people don’t make the company better — they make the founder less sharp. You need people who will push back, ask better questions, and show you what the market wants, not just what they think you want to hear.
4. Optimize ruthlessly for people you genuinely like and who love the creative process of experimenting and diligently building a company. Working with people you really enjoy gets you through the brutal slow-growth product-market-fit exploration stage. Even frustratingly slow progress is fun and stimulating when you’re with people who enjoy working together, not just those chasing startup excitement or stock options and salary. This is the real cheat code for endurance in the hard early days.
5. Despite some founders saying that starting a new company doesn’t get easier, the second startup has felt easier for me. Not because the work is easier or the path is clearer but because I have more conviction this time around. I trust the process more, I trust myself more, I trust others more, and I have a stronger sense of what’s important in the early days. I’m not discouraged by the inertia and awkward, choppy early days of 0 to 1.
6. Culture is freedom + responsibility + context. Give people the full picture of the culture you want them to co-create, real ownership, and leeway to move and exercise their own judgement. The right people usually thrive in that environment. The wrong people become obvious pretty quickly. Over time, it creates leaders who think and act like owners instead of employees.
7. Reputation compounds before revenue does, and it’s a leading indicator of future success. So staying close to customers and partners to really know their experience of the product or service is vital intelligence. Fall in love with building a systematic repeatable process for improving the customer experience. Continually revisit the customer experience. Early on, people are mostly betting on whether you’ll actually do what you say… everyone is quietly asking the same thing: can I trust this company and these founders to follow through consistently?
Exactly right. The bottleneck has never been compute or capital. Its taste and judgment about what humans actually want. Infinite compute just makes the great founders faster and the confused ones more confused. https://t.co/AmPmal8NYF
We built Luminal to be an extremely simple, powerful, and general-purpose compiler so accelerators of the future can consume models straight from PyTorch and specialize down to their exact hardware.
Chips are only as good as the software that runs them. Only one company has built a robust software stack to date: Nvidia. They've been rewarded with a $4T market cap.