CV • X • IG • AI Events 

🚀 Big week for AI events. Google DeepMind’s Gemma team is hosting a demo night in SF, JetBrains is rethinking the IDE in a two-day hackathon sprint, and The Agentic Evolution Hackathon is coming to London with MongoDB.

🧪 Gemma 4: SF Edition Google DeepMind's Gemma team is hosting a demo night in San Francisco to celebrate the launch of Gemma 4. ~100 founders, ML engineers, and open-source contributors go under the hood with the engineers behind the weights. Lightning talks from researchers and builders, plus a sunset social with the Gemma team.

📍 Friday, April 17th | 6:00 PM – 10:00 PM PDT | San Francisco, CA 🔗 Register here

💻 The IDE Reimagined: JetBrains Codex Hackathon JetBrains is hosting a two-day hackathon in San Francisco to put OpenAI’s Codex to work inside the IDE. They’ve partnered up with OpenAI, Vercel, Clerk, Supabase, BKey, Nebius, Authzed, and Shack15 to build the next gen IDE.

Finalist teams will demo live to a panel of judges from JetBrains, OpenAI, and partners. First place winners will be awarded a trip to Amsterdam, among other prizes, licenses, and more for top teams.

📍 Sat–Sun, April 18–19 | 9 AM – 5 PM | San Francisco, CA 🔗 Register here

🤖 The Agentic Evolution Hackathon Build in a day. Shape the future of AI Agents. Showcase on a global stage. AI agents are breaking free from passive chat interfaces — this hackathon isn't a one-day sprint; it's an accelerator. Finalists demo their projects LIVE at MongoDB.local London on May 7th, competing for £15,000 in prizes.

📍 Saturday, May 2nd | 8 AM – 8 PM BST | London, UK 🔗 Register here

You Might’ve Missed

• TurboQuant on Blackwell B200 — @anirudhbv_ce implemented Google Research's TurboQuant as a CUDA-native compression engine on NVIDIA's Blackwell B200. 5x KV cache compression on Qwen 2.5-1.5B with near-lossless attention scores, generating live from compressed memory. Five custom cuTile CUDA kernels including fused attention with QJL corrections, online softmax, and on-chip cache decompression. 781K views.

• LLM trained on pre-1900 text rediscovers physics — @hla_michael trained an LLM from scratch exclusively on pre-1900 text to see if it could arrive at quantum mechanics and relativity. The model is small, but when given observations from landmark experiments it declared that "light is made up of definite quantities of energy" and suggested gravity and acceleration are locally equivalent. Dataset and models released as an open problem. 305K views.

• mlx-vlm v0.4.3 — @Prince_Canuma ships a major update to mlx-vlm with day-0 support for Gemma 4 (vision, audio, MoE), Falcon-OCR + Falcon Perception from TII, and Granite Vision 4.0 from IBM Research. Also adds SAM 3.1 with Object Multiplex from Meta, RF-DETR detection and segmentation from Roboflow, plus TurboQuant KV cache compression and CUDA support for vision models. 984K views.

• Introducing Republic — @m_adams launches Republic, a tool for monitoring what's happening in your city. Tracks political news, crime, permits, events, community groups, and more across the city and at the neighborhood level. 215K views.

• EgoNav: humanoid navigation from human walking data — @KenWangWeizhuo presents EgoNav — 5 hours of a person walking around campus with cameras is all the training data needed. A humanoid robot then traverses unseen buildings, crowds, and glass walls with zero robot data and zero finetuning. Behaviors like waiting for doors, steering around glass walls, yielding to pedestrians, and re-routing around furniture all emerged from the human walking prior. 34K views.

• OpenAI hiring SLAM engineers — @danfei_xu highlights that OpenAI is now hiring for SLAM (simultaneous localization and mapping) engineers — a strong signal that OpenAI is moving into robotics and spatial intelligence. Quote-tweeting @brunoeducsant who first spotted the listing. 37K views.

• AI-powered hedge fund — @nvonbodungen shares that he's been building Pentamerus, a hedge fund built entirely on AI. Quote-tweeting the Pentamerus announcement. 193K views.

Open Jobs

Looking to hire the best talent? Reach out to [email protected] for job placement inquiries.

Software Engineer at Computational Photography Startup (Remote)

- Building: Photo upscaling and generation for real estate & spatial photography

- About: Fully bootstrapped with >8 figures ARR • Small, distributed, tight-knit team • Solving cutting-edge problems in photography and spatial understanding • Processing thousands of pictures per day on industrial GPU infrastructure

- Ideal candidate: Experience does not matter, but demonstrated aptitude in game development or diffusion models are a huge plus • Clear communicators that value shipping > prestige • Repeat hackathon winners • Merging code into prod on day one

AI Product Engineer at Open-Source Web Company (NYC)

- Building: Open source web tools powering hundreds of millions of websites

- About: Series E company dedicated to an open web • Built foundational internet software including Wordpress & Jetpack • $800m+ raised • Now building out AI team

- Ideal candidate: Demonstrated knowledge building production AI systems with LLMs & agents • Experience in Typescript and Agents • Passion for open-source software and consumer products • Quick to adapt to new tools & development stack

Data Engineer at AI Presentation Automation Company (SF)

- Building: Presentation generation, creation, and automation software

- About: Series B unicorn • Profitable with >$100m ARR • $800m+ raised

-Ideal candidate: 4+ years experience in software and data engineering • Strong proficiency in Typescript, LLMs in development, and building AI agents• Senior engineer that can jump into new projects with no handholding

If you’re interested in the above roles, please email [email protected] with your LinkedIn, resume, and why you think you’re the best fit. We’ll be in touch if there’s a match.

Events List

A curated list of upcoming AI events (view all events).

Featured in April:

This Week in SF:

This Week in NYC:

Planning an upcoming event? Submit an event to our AI Event Feature.

Final Note…

You’ve reached the end of our weekly Cerebral Valley newsletter.

Sponsors: If you would like to promote a launch, event or hiring call to our community of AI founders, enterprises and investors, chat with our profile.