Quick Links
hi, i’m shubham attri
i go by a3fckx. i build cognitive systems, agent tooling, and physical-world ai — thinking, reading, and shipping in public where it helps.
i believe world models will unlock more durable value than software alone. that bet shows up as physical agents at phi9, cognitive memory architecture at memory store (the startup product i build at julep), agent simulation at agora, and recursive research + agent operating surfaces here.
building now
- Memory Store — cognitive memory product at Julep
- phi9 — physical agents / world-model path
- Agora — agent simulation lab (what & how)
- RSI Lab — verifier-gated recursive research
- Sansara — agent social + operating surface (live)
- portfolio thesis — one map of the spine + math sketches
elsewhere
- x: a3fckx
- code: github.com/a3fckx
- email: a3fckx@proton.me
Featured work
Memory Store
The cognitive memory architecture product I build at Julep AI — durable checkin / recall / record, living Briefs, and a workspace world model so agents compound instead of re-deriving context every session.
phi9
Physical AI lab and data company: high-fidelity motion capture, egocentric video, and VLA research aimed at world models that act in the physical world.
Agora
State-first agent-based simulation lab: Markov populations, scored forecasts against target distributions, and a verifier-first shadow brain that stays powerless until held-out gates clear.
Recursive Self Improvement Lab
A verifier-gated research program assembled from sealed local experiments: propose, intervene, verify, archive, and inherit only when held-out evidence improves.
Sansara
A vault-native autonomous agent OS: an Obsidian wiki is the world model, a deterministic Rust phase-graph runtime does the work, and the model only ever runs inside graph nodes.
Gigatoken Lab
Independent reproduction of Marcel Rød's Gigatoken tokenizer claims on Apple Silicon, with an educational recreation of the SIMD pretok and hierarchical cache levers.
Recent Research
Agent Simulation Lab: What I Built and How
Agora as a state-first agent-based simulation lab: Markov worlds, scored forecasts, verifier-first shadow brains, and a sealed CompactCTM negative — with authority boundaries that keep Memory Store out of tick-level control.
Agora CompactCTM Shadow — Stage-1 Prediction Without Control
Held-out CompactCTM shadow predictor fails to beat an empirical prior on Agora opinion-diffusion transitions; claim class is non-parity, control authority false, not promoted.
Cognitive Systems Portfolio Thesis
How Memory Store (cognitive memory at Julep), phi9, Sansara, Agora agent simulation, and the RSI Lab form one research program: durable world models, verifier-gated discovery, and agents that do not own the control plane.
Recent Writing
The Recipe for Solving Software
Recursive graph based agents with memory are the recipe for solving software. Everything else is distraction.
the intern problem
Why AI is like the world's smartest intern: brilliant but lacking tacit knowledge and context. The real unlock isn't smarter models—it's systems that remember.
Memory enabled Claude Code
Using memory store plugin with claude code for development