ai-research-publications is a research portfolio, not a library, and it is built like a map

Five project directories, each with its own deep dive README

The repository is organised as a portfolio site rather than a codebase. The top level holds a README.md, a banner.gif and five project directories: ieee-aiam-2025-ga-gnn/, ijhsr-2026-celegans-mito-resilience/, stanford-urology-2026/, ucla-cosmos-2026-c1-infernotactics/ and cbe-diagram-misconceptions/.

The README states its own role plainly, calling itself the map, with every project carrying its own deep-dive README containing figures, pipelines and results. The repository map table in the README reinforces that division of labour. The IEEE project directory holds a paper/ folder, a code/ folder, an asset/ folder and a docs/ folder, and the table describes it as the full QIHGA-GNN implementation with the paper in PDF, TeX and arXiv form, SHAP figures and an acceptance certificate.

The second project, the C. elegans mitochondrial work, is laid out as notebooks/, paper/, presentations/ and private/. The name of that last folder is worth noticing, because it tells you the repository is not a single homogeneous artifact. One project is a paper with an implementation, another is a research notebook trail with a manuscript, and the Stanford project is a phase based working directory with 01-phase-1-topic-and-data/, 02-phase-2-imaging-analysis/ and docs/ inside it.

For an engineer reading this as a reference, that heterogeneity is the interesting part. Each project directory is free to be shaped around its own workflow rather than around a shared build system, which is exactly what you want in research and exactly what you do not want in a library.

Two published papers anchor a portfolio that is still moving

The README carries a research portfolio table covering 2025 and 2026 with a venue column and a status column. The first entry is the Quantum-Inspired Hybrid GA plus GNN Ensemble, published in the IEEE 7th International Conference on AI and Advanced Manufacturing (AIAM) 2025, and the focus is a quantum-inspired genetic algorithm fused with a Graph Neural Network for multimodal classification, with SHAP interpretability. The status is marked published.

The second is Mitochondrial Resilience in Wild C. elegans, published in the International Journal of High School Research, associated with the UCSB Summer Research Academy capstone. The stated method is comparative genomics of six mito quality-control genes across the CX11314 and EG4725 strains, covering selection, variant and expression analysis, and the status is also published. The README separately lists the DOI for that paper as 10.36838/IJHSR816.1.

The other three entries are live work. The Stanford Medicine project, in the Department of Urology, is listed at preliminary findings, and it is the only one described as mentored research, with kinematic ultrasound analysis of pelvic floor muscle function going from displacement to velocity to acceleration. The UCLA COSMOS 2026 entry, Cluster 1, is InfernoTactics, an actor-critic reinforcement learning agent with a CNN and MLP that dispatches firefighting resources on a terrain model of Los Angeles grounded in the January 2025 Palisades Fire, and the status is results plus a live 3D demo. The fifth, Diagrams That Mislead, is an empirical study of how simplified biology diagrams create student misconceptions, targeting CBE, Life Sciences Education, and is marked in progress.

That mix of published, preliminary and in progress is the point of the status column. It tells a reader which parts of the repository are settled results and which are still in progress.

The subject areas are deliberately far apart

The topics list is unusual for a repository: bioinformatics, biomedical imaging, c-elegans, computational biology, genetic algorithm, graph neural networks, machine learning, quantum-inspired computing, reinforcement learning, wildfire modeling, plus the programme affiliations ucla-cosmos and ucsb-sra.

The README describes the span as the full stack of research skill, moving from wet-lab-adjacent bioinformatics through clinical imaging analysis, deep RL simulation and human-subjects education research. Reading the project descriptions against that list makes the claim concrete rather than decorative. The IEEE paper is a method paper combining a genetic algorithm with a graph neural network. The IJHSR paper is comparative genomics on real strains. The Stanford project is kinematic analysis of ultrasound recordings of muscle function. The UCLA project is a reinforcement learning agent acting on a terrain model. The CBE project is an empirical study with instruments and an analysis pipeline aimed at human subjects.

For an engineer, the practical value of that breadth is as a set of worked pipelines rather than as a single method. Each directory shows the shape of a different kind of research code: a training and inference implementation, an annotated analysis notebook, a signal processing pipeline over imaging data, a simulation with an interactive demo, and a study with instruments.

It also sets a boundary. There is no shared library here. The genetic algorithm in the IEEE project and the actor-critic agent in the UCLA project share no code, and nothing in the repository map suggests a common package that a reader could depend on.

Affiliations and credentials are part of the repository's content

The README carries a set of badges and a highlights table that mix research outputs with academic and competition results, and a reviewer has to read them as claims about the person behind the code rather than as properties of a tool.

The research side lists lead authorship of the IEEE AIAM 2025 paper with the identifier AIAM-6203, lead authorship of the IJHSR 2026 paper, a place as research scholar at Stanford University School of Medicine in the Department of Urology, and selection for UCLA COSMOS 2026 Cluster 1. The academic side lists dual perfect scores at USACO 2025 US Open, a national perfect score at AMC 10 and 12, and founder and president roles at AI Ethos, Inc., a 501(c)(3), and NextGenAI International, described as ethical-AI education for more than 75 students across three countries.

One figure in the highlights is a statistical claim worth handling carefully. The README says fewer than 1 percent of U.S. high schoolers publish in AI or computational biology venues, and that the repository holds two such papers from a single sophomore year. There is no citation attached to that percentage in the README, so treat it as the author's framing rather than a verified statistic.

What is verifiable from the repository itself is narrower and more useful: two accepted publications with venue names and a DOI, and directory listings that show papers, notebooks, code and figures for each. The affiliations are the ones the author lists, and the homepage in the repository listing points to a personal site and a Google Scholar profile that the README also links.

No package, no releases, and a licence you have to read twice

This is not a library and it does not pretend to be. The primary language of the repository is listed as HTML, the repository has no GitHub releases, and there is no build script, no package manifest and no dependency list. There is nothing to install and nothing to import.

The licence story needs care. The repository listing shows no licence identifier, and the top level entries are the README, the banner and the five project directories, with no LICENSE file at the root. The README, however, carries a badge reading MIT plus per-project, and a navigation entry points to a licence and data use section further down the document. In other words, the badge asserts a per project licensing model while the root of the repository ships no licence file to read.

That is not an unusual arrangement for a portfolio with several separately licensed sub projects, but it means the terms are per project and the ones you care about are in the individual directories. Do not infer the terms for one project from the badge.

Maintenance status has the same shape. The last push was on 2026-09-20 and the repository is not archived, so the work is current, but with no release tags there is no version to pin and no changelog to diff, which is normal for a portfolio and also a reason it cannot be consumed as a dependency.

Where this repository is the wrong thing to reach for

The honest limitation is that almost everything in here is unusable as a building block. There is no package to add to a project, no documented API, no test suite, and no versioned releases. If your goal is to run a quantum-inspired genetic algorithm with a graph neural network, or to reproduce the wildfire response agent, you are reading code for ideas rather than consuming an artifact, and you should expect to do the environment work yourself.

There is also a visibility limit. The map shows a private/ directory inside the C. elegans project, which is not readable from outside, and the README says the final deep dive READMEs with figures, pipelines and results live inside each project, so the details of the pipelines and the results are only partly visible from the top level. The description also mentions biomedical imaging and science education research among the active programs, and the depth of those two lines of work is not visible in the repository listing at all.

The alternative, if what you want is the ideas rather than the repository, is to go to the artefacts themselves. The paper folder of the IEEE project holds the paper in PDF and TeX, and the IJHSR project has a published manuscript and a DOI, and the Google Scholar profile linked from the README collects the citation record. A researcher who wants to cite or check the work starts there, not with the directory listing.

For a student or a small team wanting a template for the same kind of portfolio, the layout is the transferable part: one directory per project, each with a paper folder, an analysis folder, an assets folder and a docs folder, plus a top level README that acts as an index with a status column.

Editorial conclusion

Read this repository if you want a concrete example of how a small research group organises code, figures and manuscripts for several projects at once, particularly the layout of the published paper directories and the separation of notebooks, analysis and assets. Do not adopt it as a dependency, since it ships no package, no release tags and no root licence file. Verify first which directories you actually need, since the repository map shows a private/ folder inside the IJHSR project that will not be readable, and check the per project licence terms yourself, because the repository listing shows no licence identifier while the README badge claims MIT plus per-project terms.

Frequently asked questions

Official sources

1. aaryansamanta/ai-research-publications on GitHub
2. Issues
3. Project website
4. README