THINKING
MACHINES
A growing group of NCSU students. We train deep neural nets on large datasets and meet weekly to learn, build, and push the state of AI forward.
How We Think
Understanding neural systems by building them.
We learn backpropagation, train on GPUs, benchmark performance, and experiment to understand how models actually work.
We believe the best way to learn AI is by building it, one weight update, one tokenizer, one training run at a time.
Events
Fall ’26
A semester of building, in public.
- Week01
Introductions
Meet the officers, see what we'll build this semester, and run through deep-learning fundamentals.
- Week02
Running models on GPUs
Guest talk from NVIDIA on how inference actually works on real hardware.
- Week03
Building GPT-2 with nanochat
Walking through Andrej Karpathy's nanochat: tokenization, training loop, sampling.
- Week04
Backprop, by hand
Trace gradients through a small network on paper, then verify in PyTorch.
- Week05
Benchmarks & beyond
How modern models are evaluated, and where the metrics quietly lie.
What We Do
01
AI News & Papers
Discuss the latest advances and key research through curated paper breakdowns.
02
Competitive Projects
Build and iterate on AI and systems projects, including competitions.
03
Concept Deep Dives
Explore core ideas and model internals: backprop, GPUs, training, system design.
04
Workshops & Speakers
Hands-on workshops and guest speakers from industry and research.
05
Community & Game Nights
Build friendships and have some fun with people who care about the same things.
About Us
Thinking Machines is a growing group of NCSU students passionate about artificial intelligence and systems.
We meet every Friday at 5pm at the NC State Albright Entrepreneurship Garage.
We host weekly meetings during the semester to learn, build, and explore together.
Open to all NCSU students who love to learn and build.
Fall ’26Manifesto
“What I cannot create, I do not understand.”
