Modal

Serverless in SF 📍

Helping companies like Firebender create new AI agent benchmarks in 1/100th of the time 🚤 Read their full blog post on how they built a benchmark specifically for Android tasks, forked from the popular SWE-bench repo: https://lnkd.in/gNKpfPHu

If you missed it, we had an amazing IRL meetup with the MLOps Community and Google at the Google Cambridge office yesterday! Incredible turnout, something we'll be doing again soon. A huge thank you to Claritza N. Abreu for helping host the event, and our main speakers Raquel Harrison and Alex Burdenko. I gave a small talk as well, showing off how well Modal and Boundary (YC W23) BAML work together. We [live!] deployed Llama-3.2-11b and used it to extract a few different PDF files into structured, type-safe JSON. You can find that repo here: https://lnkd.in/e-EA-uKs If you want to be at the next meetup, be sure to join the MLOps Community slack and the Boston channel! https://lnkd.in/esGiPwW7

Going off the beaten path and building your own AI stack? ✅ More control ✅ More customization ✅ More privacy But when your app starts gaining traction, infrastructure can become a nightmare—especially compute. In this walkthrough, I deploy my own AI chat API (powered by LLaMA 3.1 Instruct) using Modal’s serverless GPU infra. Then I put it to the test: how does it scale when things go viral? If you’re thinking about self-hosting an LLM or building OpenAI-compatible APIs yourself, this might be the guide you’ve been looking for. Links in the comments 👇

Still buzzing from last night’s Growth Engineering Demo Night at South Park Commons with four of NYC’s fastest-growing tech companies: Profound, Modal, Clay , and Hebbia! Huge thanks to our incredible demo'ers: Charles Zhou (Profound) showed how brands can understand their presence in AI search. Akshat Bubna (Co-founder, Modal) demo’d DoppelBot, which can clone anyone in your Slack. Adithya Ramanathan (Head of Applied AI, Hebbia) took us through an AI-powered analysis of the JFK assassination documents. Everett Berry (Head of Growth, Clay) identified customers live in the audience and used AI to auto-generate personalized pitches. What a night! ⚡️

Your self-hosted chatbot just went viral—can it handle the traffic? We deployed an OpenAI-compatible chat API using Meta’s LLaMA 3.1-8B (quantized) on Modal. Then we hit it with 1,500 (!) simulated users. No Docker, no Kubernetes—just Python and Modal’s GPU serverless infrastructure scaling in real time 🤯

🏆 We’re honored to have been voted # 2 most promising early-stage company on the 2025 Enterprise Tech 30 list. This is the 3rd year we’ve made an appearance on this list, which crowdsources recommendations from 100+ VCs. We’re in good company, too, with some our customers: Ramp, Scale AI, Cognition, Anysphere, Cartesia Join us before we graduate out of the early-stage category 💅 modal.com/careers Check out the full Enterprise Tech 30 list from Wing Venture Capital and Newcomer here: https://lnkd.in/eCsWmSC

Thrilled to announce that Everett Berry from Clay is now joining us on March 27th at South Park Commons NYC to chat all things growth hacking 📈 Joining a stacked lineup: - Our very own Akshat Bubna - Dylan Babbs from Profound - Adithya Ramanathan from Hebbia Thanks again to Dylan Itzikowitz for making this happen. This is turning into an event you won't want to miss!

🌟 Here's what's new in Modal this month! 🌟 🤑 Price drops 🐍 Conda-forge availability 📽️ New image-to-video example 📊 Guide on GPU utilization 🛣️ Guide on API platforms vs self-deployment 🎁 New logo plus billboards in SF! https://lnkd.in/erv9UDu5

We're hiring across all our offices! 🗽🌉🇸🇪 Apply at modal.com/careers