Adrian Michalski

The Gemini 3 Hackathon powered by @GoogleDeepMind is coming to an end. One day left and I’m sharing my demo today 🥰 Tomorrow: final tweaks and submission on @devpost Enjoy Blasto - play, experiment, and learn AI engineering through an adaptive teaching system that accompanies you throughout the game.

@ChomatekLukasz Miesięcznie z 70-80% planu Max 100$ na CC. Gdyby to leciało przez API to pewnie Jensen nie byłby zaniepokojony 😆

„Pozwólcie, że przedstawię wam eksperyment myślowy. Załóżmy, że macie inżyniera oprogramowania lub badacza sztucznej inteligencji i płacicie mu 500 000 dolarów rocznie” – powiedział Jensen Huang, prezes Nvidii. Dodał również: „Pod koniec roku zapytam go, ile wydał w tokenach. A jeśli ta osoba powie 5000 dolarów, to sygnał dla mnie, że coś jest bardzo nie tak. Jeśli ten inżynier, który wydał 500 000 dolarów, nie wyda tokenów o wartości co najmniej 250 000 dolarów, będę bardzo zaniepokojony”. A ile wy miesięcznie wydajecie na tokeny?

@benjaminakar @darrenjr Last 2 projects 🤌 blasto.xyz silene.systems

we're hiring builders. don't send a cv. send what you’ve shipped. projects. products. design. github. show taste. show ownership. late nights. early mornings. you only get a few real shots. if you believe in what we're building, message me. remote in europe + north america. bonus if you grew up on minecraft.

@annawitten @JakubStyczynski Niesamowite. Startupy walczą przez lata o możliwość współpracy z ochroną zdrowia, aby wspólnie rozwijać i testować rozwiązania AI. Tymczasem w przetargu startują firmy, które nie potrafią nawet dobrze przefiltrować katalogu na HuggingFace i sprzedają dostęp do modeli open source

Co tu się wydarzyło 😳 Kuriozalny przetarg finansowany z KPO opisuje @JakubStyczynski. O tym, że wygrała go konkretna firma zdecydował fakt, że oferowała modele AI do wykrywania Harrego Pottera i generowania anime-pornosów. zero.pl/news/spor-o-sy…

@cerebral_valley @Meta @nvidia @SHACK15sf More online event guys 🙏

Do you want to demo your project at the Meta booth during GTC? 😎 Join @Meta and @nvidia, in partnership with CV, for a full-day hackathon at @SHACK15sf, writing high-performance GPU kernels with Helion, PyTorch's new kernel authoring DSL that delivers higher performance in fewer lines of code with autotuning. 📅 March 14th — Right before NVIDIA GTC. The perfect warm-up. 🏆 Prizes & perks: > Nvidia GPUs and Nvidia DGX Spark > Demo your project at the Meta booth during GTC > GTC conference passes > Ray-Ban Meta glasses > Mentoring from Meta AI researchers & NVIDIA engineers 📍 Fully in-person | Teams of up to 4 | Rolling review, limited spots Register below 👇

Even if this was ultimately a marketing play - I think Claude Code with Opus 4.6 is a genuinely great tool. Domain experts with a technical inclination will build software that solves their own problems. Whether that turns into real products comes down to the individual's commitment and focus. As for me, my main takeaway from the whole Claude Code Hackathon just arrived - a swag cap with Clawd on it. So I guess I'm doing my part for Anthropic's marketing too.

Anthropic is certainly aware of this. On the day the results were announced, they posted a thread sharing stats on "In what domains are agents deployed," writing: "Software engineering makes up ~50% of agentic tool calls on our API, but we see emerging use in other industries."

Was the Claude Code Hackathon a marketing masterpiece or has Opus 4.6 broken down the barrier to building software for everyone? Only 500 people participated (out of 13,000 applicants). Compare that to Bolt New's "World's Largest Hackathon" — 128,460 applicants. Or Google's Gemini 3 hackathon — 35,000.

Thank you for that framing. I've spent some time working on CDSS for rare diseases and a few weeks ago started building a platform that extracts and structures patient-level data from published case reports at scale. Trying to turn scattered clinical evidence into something usable for computational phenotyping. Reading your vapor phase description felt like seeing the theoretical frame for what I've been doing on intuition. Case reports are not EMR, not biobank, and obviously biased, but maybe well-curated edge cases are exactly what's needed to map the phenotypic extremes.

Phenotypes are like Water | Beyond the Ion Channel For #RareDiseaseDay 2026, I revisited a 2023 idea: phenotypes exist as ...ice (trial readiness) ...water (real-world data) ...and vapor (computational phenotyping) epilepsygenetics.blog/2026/02/28/phe…

dataset is available here zenodo.org/records/187990…

Three days ago, on #RareDiseaseDay, we released a structured clinical dataset for Fabry disease, extracted from published case reports using our AI pipeline. I spent the last 3 weeks building an agentic AI system that extracts structured patient data from rare disease case reports. It started at a Claude Code hackathon and grew from there. This first version of the Fabry dataset includes 154 patients from 93 publications, over 2,000 symptom observations mapped to HPO, 1,264 lab measurements, and 60 unique genetic variants. All available as CSV files under CC BY 4.0. In upcoming releases, we plan to expand the dataset with more patients and publish a paper describing the agentic extraction pipeline. You can find more details about the dataset in the carousel below, and the full description on Zenodo. If you'd like to explore the data interactively, head over to Silene Systems. We also have datasets for other rare diseases on the platform - not yet fully verified, but available to explore and download.

@tunguz The difference is that when you show a client the first AI demo, they don't understand why you need to work on it for another three months. After all, you already have turbo Simpson 🥲

Seems about right

@JorgeCastilloPr 😅

Me when Claude already wrote like 3k lines of code and I notice an error on my prompt

@pattybuilds Hard to believe it’s already been a year since PMF or Die 😯

first time in Okinawa | 初めての沖縄

@Michaelzsguo @claudeai Yes ->

@claudeai So the winners of the Claude Code hackathon were: - a personal injury attorney - an interventional cardiologist - an electronic musician - an infrastructure/roads systems worker - and one software engineer That should tell you something.

Our latest Claude Code hackathon is officially a wrap. 500 builders spent a week exploring what they could do with Opus 4.6 and Claude Code. Meet the winners:

@kawecki_maciej Brawo dla Michała, zarówno za projekt jak i poświęcenie, żeby polecieć do SF. W Hackatonie brało nas jeszcze 2 oprócz Michała. Podsyłam wszystkie projekty dla ciekawych: postvisit.ai silene.systems revlo.review

@dalraejin @AnthropicAI @cerebral_valley Hey 👋

@M_Adrian2 @AnthropicAI @cerebral_valley Hi Adrian, nice to know you here!

@ZehnerLukasz @AnthropicAI @cerebral_valley Haha thanks 😅

@M_Adrian2 @AnthropicAI @cerebral_valley You're a beast! Hackatony badzcie w opiece ;-)