South Africa PhD in AI from the University of Cape Town 5+ Years as Machine Learning Research Engineer AI Safety
Cape Town
Joined July 2022
- Tweets190
- Following218
- Followers79
- Likes7.5K
This is awesome!
Hey @dwarkesh_sp!
@NomadsVagabonds @gilg_oscar @TomTaylor256 and I already did this and had similar findings.
nitter.gq/NomadsVagabonds/status…
Claude Formanek retweeted
Excited to announce the release of TestBED, an open-source toolkit for Bayesian Experimental Design.
TestBED provides the core primitives for building BED algorithms, alongside reference implementations of established methods.
Available now on PyPI: pypi.org/project/torch-testb…
I'm delighted to announce the release of TestBED, an open-source toolkit for Bayesian experimental design. It provides a flexible and extendable foundation for coding adaptive design methods and quickly benchmarking them
arxiv.org/abs/2611.06633
github.com/testbed-benchmark…
Claude Formanek retweeted
Gemini 4 Argon is now globally available.
Built to solve your hardest problems from start to finish, the model brings frontier capabilities across coding, enterprise workflows, and cyber defense.
What a good time! 🌞🤖💚
Lecture recordings and a blogpost recapping the Summer School will be released soon. Sign up to our mailing list to learn about future events: cooperativeai.us1.list-manag…
Best Poster Award 🥇
Work done together with @KaliTessera and @hedman_marcel 🚀
We also announced the Best Poster Awards from our Summer School poster sessions. Congratulations to @ClaudeFormanek, whose poster on ‘Molten Pot: Evaluations & Datasets for Social Offline Reinforcement Learning’ received the top prize.
Claude Formanek retweeted
During our Cooperative AI Summer School in Toronto this month, participants developed project proposals around cooperative AI research. They focused on questions such as how secretly loyal AI would act in multi-agent systems and how well LLMs can extrapolate human preferences when acting or voting on someone’s behalf.
Claude Formanek retweeted
Evidence that South African special forces murdered top detective shared with BBC Africa Eye.
bbc.in/4z6NbIR
Claude Formanek retweeted
Can AI agents learn what you think and represent you in a debate you never attended? In today’s blogpost, @Jolow99 and @Oscarduys share their thought process and learnings after building Habermolt: the first public platform for AI-delegated deliberation.
This looks like a really good framework! I am a big proponent for more Multi-Agent Safety Evals. The rate at which agent's are being deployed into the wild by individuals and organisations means that empirically quantifying multi-agent risks should be an urgent priority!
Today we're releasing Orbit v0: a framework for multi-agent safety and security research.
If you've tried running multi-agent experiments, you know the drill - weeks of infra before your first result. Built on Inspect, Orbit lets you swap environments, threat models, and topologies without rebuilding anything.
Claude Formanek retweeted
Most AI evaluations focus on an individual agent, even though many key threat models and safety-critical domains involve multiple interacting agents. We’re excited to be supporting Orbit, a new framework (built on Inspect) to help researchers address this gap.
Today we're releasing Orbit v0: a framework for multi-agent safety and security research.
If you've tried running multi-agent experiments, you know the drill - weeks of infra before your first result. Built on Inspect, Orbit lets you swap environments, threat models, and topologies without rebuilding anything.
This is awesome!
Can LLM agents coordinate in long-horizon, open-ended worlds?
We evaluate 13 modern LLMs in a new benchmark where agents must work together to explore, communicate, trade resources, craft tools, build structures, and fight mobs.
TL;DR: Most agents struggle, averaging only ~6% normalised return. Yet on the hardest setting, zero-shot Gemini 3.1 Pro performs comparably to the best MARL agent trained for 1 billion environment steps.
More broadly, we find coordination is a distinct bottleneck beyond long-horizon task competence, with communication having the largest effect in our harness ablations. 🧵👇
Just gave a presentation at an #ICML workshop on our Molten Pot: Evaluations and Datasets for Social Offline Reinforcement Learning! 🤖💬🤖Really happy that this paper received the Runner-up Best Paper Award 🥈
And congrats to co-lead author @hedman_marcel 🚀
Claude Formanek retweeted
Some initial takeaways from our pre-@icmlconf workshop hosted in Seoul with @foresightinst on ‘Supercooperation: The Future of AI for Democracy’, exploring how AI tools for collective deliberation connect to near-term democratic challenges and longer-term risks of power concentration:
Wow, amazing work Dries 👏🇿🇦
ARC-AGI-3 is built different, it has dumbfounded almost all regular attempts so far because it's so much harder than anything that came before. It has no rules, it's agentic and has no explicit goals, they need to be discovered.
@tufalabs won the first milestone of @arcprize
> There is no language built into the benchmark, but these guys "put the language back in", because in their view - it's the best way to climb up the notional "abstraction mountain" and effectively use many of the abstractions which have evolved over millions of years of language evolution.
> They built a novel harness "The Duck" around a 27B open weights model (Qwen 3.6) to solve extremely challenging and novel reasoning problems that require abstraction.
> This is the launch video of their winning agentic harness, "The Duck". We have also released an exclusive interview with them on MLST, just dropped.
> The million dollar question is: what will @fchollet think about how they've done it, and is this a step towards AGI?
Claude Formanek retweeted
20-year-old centre-back Mbekezeli Mbokazi has been a bright spot for South Africa.
His 86.0 Overall Grade ranks 1st among all CBs (so far), with excellent grades in Positioning (86.6), Clearance (82.7) and Passing (76.7). 🇿🇦
| @ChicagoFire #BafanaBafana |
Claude Formanek retweeted
More than just a goal 🤩🇿🇦
Look at the celebrations as South Africa book their place in the Round of 32 ⏭️
#SSFootball | #DStvFIFAWorldCup2026 | #FIFAWorldCup
Well done!! 🇿🇦♥️
𝗦𝗶𝗹𝘃𝗲𝗿 𝗺𝗲𝗱𝗮𝗹
An incredible week for a team that was not ranked in the top 3 before the tournament begins to fall just short of the gold. We are proud of this team, and proud of the support they received.
Its always about #MoreThanHockey
