
Maker at @h2oai, proud father of twins, maintainer of Sparkling Water, contributing into open source AI projects, and still believing that coding is beautiful!
Joined March 2008
- Tweets3.7K
- Following3.8K
- Followers762
- Likes1.6K
Pinned Tweet
The Wildfire Challenge is opened for submissions! Do not hesitate, register at h2o.ai/wildfire/ and submit your AI App to win prizes 🏆
And if you have any questions visit H2O.ai community slack h2oai-community.slack.com or forum discuss.challenge.h2o.ai.
1/5 🏆 TimeH2O ranks #1 on EnergyBench, a live benchmark for day-ahead energy forecasting!
TimeH2O currently leads the Elo leaderboard, ahead of competing models including TimesFM, Chronos, and Timestral.
🔗 energybench.foretec.co/
Today at #SFF2026 UOB's Dongtao Li shows how the bank built MAP, one platform for every model it ships, with Vincent Lam of @h2oai.
📍 UOB Pavilion Mini Stage
🗓 1:15 to 2:00 PM
#AIinBanking @sgfintechfest
TimeH2O Outperforms Mistral, Google, and Amazon in Time-Series Forecasting!
We're proud to introduce TimeH2O-v1, the #1 open-weight time-series foundation model, ranking first among 32 models on the fev-bench leaderboard across 100 forecasting tasks.
#TimeSeries #Forecasting #Fevbench
A 4-billion-parameter open-weight model just beat Microsoft-Decision-1 and OpenAI's Decisions API on JevBench. Faster and more capable than both.
That model? H2O-Lightning-4B under Apache 2.0 ⚡
It's also faster than Jev and nearly as capable while also supporting images.
#OpenSource #EnterpriseAI #Jevbenchmark
A new paper from Apple engineers ends with a line we couldn’t have written better ourselves:
"if idle macOS devices across organizations [...] can be securely federated, they could form a massive, shared GPU resource pool that benefits the broader ecosystem."
That’s what we’re building with Darkbloom.
The paper's Minions is a system that pool idle Apple Silicon Macs into a distributed GPU cloud for on-prem inference. Queue times went from 10+ hours to minutes. Zero new GPU purchases. Zero infrastructure cost.
It runs inside Apple, on coworkers’ Macs, using the company’s own storage and access controls. When a team needs more capacity, they Slack colleagues and ask them to turn on the executor. Even in that setting, the top adoption concern was data confidentiality. People were wary of their jobs running on a colleague’s Mac. Apple’s fix was to keep each team’s work on its own team’s machines.
Which brings it back to that one word in the conclusion: “securely.”
Inside a company, you can shrink the network until you trust everyone in it. On an open network, the machine belongs to a stranger and there’s no one to Slack. Keeping prompts private and outputs honest on hardware you don’t control is the hard problem that we solved at Darkbloom.
Apple built the on-prem version.
Darkbloom is the open one.
Michal Malohlava retweeted
Another sovereign AI partnership landed this week.
H2O.ai and CAN.B Group teamed up to bring governed, sovereign AI to Australian government agencies, critical infrastructure, and regulated industries. On-prem, sovereign cloud, air-gapped. Full model governance and auditability.
When a company that has raised $256M and serves more than half the Fortune 500 makes sovereignty its headline, the category isn't a niche anymore. It's where enterprise AI is heading.
We've been building toward the same thing from the other end of the market.
Regulated institutions were told to adopt AI and forbidden from sending their data out to do it. The big platforms are answering with flexible hosting, model portability, and governance layers. All real, all necessary.
But there's a layer underneath that most sovereignty stories still skip.
You can put the model in a VPC. You can host it on-prem or air-gapped. The harder question is the retrieval substrate, the full index of everything your company knows, plus every query and trace it generates. That's where the sensitive surface actually lives, and it's the part that most often still sits somewhere the customer doesn't fully control.
Model isolation is not knowledge isolation.
That's the gap we've spent the last four months building @enclavelabsai to close. Sovereign retrieval that runs inside your own cloud, so your index and your data never leave your walls. Sovereignty you can verify in your own logs, not read in a contract.
The demo is ready, and we're now looking for early pilot partners. If you're inside a regulated institution wrestling with exactly this, and you want to try sovereign retrieval in your own environment, let's talk.
Cal: cal.com/shashank-bhardwaj-fw…
hashtag#DataSovereignty hashtag#EnterpriseAI hashtag#AIInfrastructure hashtag#RegTech
Michal Malohlava retweeted
H2O.ai Joins Open Secure AI Alliance to Strengthen AI Software and Agents Security ow.ly/fnzT50ZyROC #TechnologyNews #AI #TechNews #CIOCommunity #CIOLeadership #CIOInfluence #TechLeadership #ITStrategy #FutureOfIT #TechTrends
Michal Malohlava retweeted
Introducing the Science One Framework, an experimental autonomous research prototype that builds verifiable evidence chains. We demonstrate how natively maintaining these chains eliminates hallucinated citations and enables reproducible AI science. Read the blog to learn more: goo.gle/4wzlGWU
I feel like living in the future every day. This is another example to try: Tau Robotics (@taurobots): a humanoid robot that cleans your home, one hour for $30. Now, i am waiting in waitlist 🦾, you can too tau-robotics.com/waitlist?re…
3
229
Michal Malohlava retweeted
Introducing autoresearch for arXiv papers
Change 'arxiv' to 'autoarxiv' in any paper URL
An agent deploys to resolve setup issues on the codebase, run a minimal reproduction, and estimate full replication cost. Read more below
Michal Malohlava retweeted
I wanted to share a bunch of my favorite hidden and under-utilized features in Claude Code. I'll focus on the ones I use the most.
Here goes.
Michal Malohlava retweeted
Introducing TurboQuant: Our new compression algorithm that reduces LLM key-value cache memory by at least 6x and delivers up to 8x speedup, all with zero accuracy loss, redefining AI efficiency. Read the blog to learn how it achieves these results: goo.gle/4bsq2qI
Michal Malohlava retweeted
GUI for fine-tuning LLMs without code
github.com/h2oai/h2o-llmstud…
Big momentum at #NVIDIAGTC 🎉
During his keynote, @nvidia CEO Jensen Huang highlighted the rise of AI-native companies — and H2O.ai was featured on the slide.
Proud moment to be recognized alongside leading innovators in the ecosystem 🦾
#AINative #H2OSuperAgent #EnterpriseAI


























