| Machine Learning | 62 |
|---|---|
| Computer Vision | 43 |
| Data Science | 38 |
| NLP & LLMs | 70 |
| AI Agents | 66 |
| MLOps | 32 |
Each node is one domain, scored only from evidence in these public repositories. Select one to see why.
Deep, repeated, well-engineered work in this domain.
28 public repositories → 14 with AI work, evaluated. The other 14 (repositories without AI code, forks and empty repositories) are not scored and have no effect on the result. How repositories are selected
Ordered by how much each project adds to the Atlas score, not by its own score. A project with evidence in several domains can rank above one with a slightly higher project score. Choose a domain to focus on the projects with meaningful evidence there.
22% of Atlas Score
kvidcache is a research repository that investigates temporal redundancy in transformer KV caches by implementing and evaluating various predictors and codecs for residual compression.
16% of Atlas Score
Sydes is a cost-aware coding agent control layer that reduces LLM spending by combining local repository intelligence with adaptive model routing, budget-aware control, and verification feedback. It is built on top of the Pi coding agent and provides a deterministic control boundary for agent execution.
12% of Atlas Score
This repository implements Jordan-Spectral Attention (JSA), a spectral-shift replacement for Transformer self-attention in language modeling. It includes training scripts for baseline and JSA models, along with experimental results showing JSA's effectiveness in reducing perplexity with fewer parameters.
11% of Atlas Score
The repository provides an end-to-end workflow for parametric L-bracket design, using C#/PicoGK to generate a dataset, training surrogate regression models (Random Forest, MLP) in Python, and applying Optuna to minimize predicted mass via generative design.
8% of Atlas Score
mgvs is a research/competition codebase for Markov-Guided Verified Search, a solver that uses LLMs to propose bounded actions on a symbolic reasoning state with verification and beam search.
7% of Atlas Score
BioAgents is a Python framework for building multi-agent reasoning systems where specialized agents generate, debate, and refine hypotheses through a swarm-inspired process, with example applications in code review and travel planning.
