Focus
Choose fewer things, define them sharply, and carry them far enough to matter.
Rough Idea
Manhattan summer treated as a creative residency, research lab, and body-of-work generator.
This is a personal AI/ML studio season. The point is not just to spend the summer learning or building, but to deliberately convert curiosity into visible artifacts with technical depth, strong taste, and enough rigor that the work can be evaluated by serious people.
The throughline is repeated deep work, a clear record of progress, and a portfolio of outputs that proves ability rather than merely describing it.
The design is meant to feel connected to New York: direct, dense, exposed, and in motion. Brutalism fits because the thought process, progress, structure, and work should stay visible.
125
days of focused studio work
10
core themes across AI, science, systems, and society
12
linked repos tracked as one work system
Themes
The recurring lenses for selecting sessions, projects, research sprints, writing, and public artifacts.
- Theory and foundations Math, stats, theory, and first-principles AI understanding.
- Art and creativity Generative tools, taste, expression, and creative practice.
- Applied science: applied ML, deep learning, and AI Useful AI systems shaped by constraints, experiments, and evals.
- Social good + Effective Altruism Data science and AI for specific problems that help people.
- Applied frontiers and research Frontier sprints that produce code, findings, and limits.
- AI and human-centered problem discovery and solving People, workflows, incentives, and decisions before models.
- Systems for the modern AI era Agents, evals, data flows, inference, and iteration loops.
- Data science + X + research Evidence, causality, statistics, and domain depth.
- AI + X + research Intersections where AI changes what can be built or studied.
- Philosophy, economics, prosperity + AI Intelligence, markets, abundance, values, and progress.
Implementation Stance
Everything here should be implemented neatly, with the standard visible in the artifacts.
Craft
Make the work legible, reproducible, evaluated, and pleasant to inspect.
Need
Start from real problems, real questions, and real constraints.
Taste
Care about judgment, framing, quality, and the difference between finished and inevitable.
Anchor Outcomes
The season should produce concrete evidence, not only intention.
YC-level product-building project
One serious company-building / FOSS project with a real user problem, sharp wedge, working system, evaluation, and public artifact.
Research outcome
One serious research paper, plus preprint or workshop-grade outputs where possible.
Required Repo Map
One shareable control panel for planning, contribution, tracking, and follow-through across the wider 2026 work system.
nthExperiment
An 8-month self-directed diploma on becoming a polymath in the AI era.
theFalltoRISE
A project lab for understanding important problems as dynamic systems.
Thursday-Learning-Hours
A filesystem-first archive for disciplined technical sessions.
how-to-do-ai-research-2026
An open living lab for serious AI/ML/DS research engineering.
AI-Science
AI systems for scientific experiments, authorship, and review.
FAIRE
Frontiers in AI Research and Engineering through sprint-based experiments.
ai-first-editorial
AI-first stories, insights, essays, and narrative artifacts.
aops-fms
Individual-scale frontier model research engineering on open-weight models.
mlbrix
Reusable ML bricks, utilities, components, and experiment infrastructure.
advanced-deeplearning
Advanced deep learning architecture, training, inference, and analysis.
Convergence-2026
A convergence map across AI, inference, interpretability, and learning.
applied-ai-2026
A problem-first applied AI course and compendium for 2026.