Pinned
Eli Mernit
1,046 posts
Eli Mernit
@mernit
The plan: - Get accepted to YC W25 - Launch an AI agent app, hard pivot after 1.5 weeks - 2 days of soul searching, abyss of sorrow - Meet your group partner for office hours, possibly the most enlightening 20 minutes of your life - Reach out to engineering manager from- The plan: - Graduate from YC X25 - Prepare for fundraising at demo day, $2M on $20M post - 12K MRR growing 80% MoM - $275k committed from rich family friend and random founder you met in an Uber one time - Meet with prestigious seed fund, setup time for second call with
- SF tech meetup starter pack - 98% dudes - Hosted at office of Series B tech company - Attendees are founders trying to sell you their product or people looking for jobs - 25 minute talk on esoteric tech topic - Send LinkedIn connection requests to 5-10 people you’ll never
- I was fired from OpenAI tonight. I was in charge of SRE on the weeknight rotation. It was a fun first job out of college, and I was excited to leverage Devin and Claude to write Terraform policies for our DNS records. If you know anyone hiring SREs, let me know. TC: $415K
- Remember when people debated whether to run Jupyter notebooks in production? Feels like a decade ago
- Things that are in: - Talking to users - Zone 2 cardio - Raising prices - “Founder mode” - Making copy changes based on customer anecdotes and vibes - SF - Going on camping trips with your co-founder - Getting default alive Things that are out: - Hiring salespeople pre-PMF -
- I didn’t win the Olympiad I worked at the bike shop fixing flat tires I didn’t graduate from Stanford I studied philosophy at a state school I barely got into I didn’t work at OpenAI I worked at a startup no one heard of and built apps after work We’re not “cracked”If teams have IMO golds, they report on those. If they don't have IMO, they report on ex-OpenAI engs If they don’t have ex-OpenAI engs, they report on "we're cracked" If they’re not cracked, they report on ‘raised from t-1s’ If they don't have $, they report on advisors.
- Replying to @forgebitzI will always appreciate the user who asked me if the reason his requests were slow was because we ran out of uuids
- If you're using serverless, latency is pretty important. Especially for GPU workloads. Just shipped a little addition to @beam_cloud that shows the full breakdown of a serverless cold start Should make it easier to debug your functions and achieve the lowest latency
- PromptLayer is such a powerful tool in the LLM stack. I'm loving the experience so far!Great article by @beam_cloud on how to use @langchain and @promptlayer with their platform! docs.beam.cloud/getting-starte…
- Our serverless workflow at @beam_cloud is pretty fast. Takes only ~12 seconds to deploy a web endpoint which includes authentication and will autoscale with traffic.
00:00 - @beam_cloud is now powering a one-of-a-kind generative AI experience for Coca-Cola ✨ Earlier this month, Coca-Cola dropped a new limited-edition flavor called K-Wave. Each can has a QR code — when you scan it, you'll get dropped into an immersive AI experience, powered by
