Open projects

10 available · worldwide
AllAIML / DataOpen SourceWeb devDesignMarketingContent
TypeScriptfeaturedOpen

Someone tried to build AGI. Pick up where they left off.

Someone tried to build AGI with microservices. They got halfway. We have [the repo](https://github.com/andrewrgarcia/this-will-change-everything). It has a cognition engine, a memory layer, a reflex loop, and an event bus. It almost does something. There's even a [frontend](https://this-will-change-everything.vercel.app/) — signals, nodes, a P=NP claim, and a button that doesn't quite work. Start from it. Take it somewhere real. You decide what it becomes — but it has to run, and it has to grow from what's already there. See the [README](https://github.com/andrewrgarcia/this-will-change-everything/blob/main/README.md) for the full brief.

TypeScriptNext.jsOpen SourceArchitectureAIWeb dev~2 weeksremote
AG
Andrew Garcia
Remote
Apply →
tradingfeaturedOpen

C++ Trading Engine — ongoing, team welcome (open source on completion)

## C++ trading engine — resolve all marked TODOs. I will provide a seed repo: working build, passing tests, clean C++20. The system runs signal → order → fill → PnL end-to-end. Ten TODOs across four areas: ### Signals - VAR(p) — Eigen OLS skeleton in place, algorithm documented ### Execution - Resting limit orders with price-time priority - Persistent order state (NEW → PARTIAL → FILLED → CANCELLED) - Partial fills across book updates - Nonlinear slippage (square-root impact, volatility dependence) ### Analytics - Position-aware PnL and trade pairing (open ↔ close) - Turnover, Sortino, volatility metrics ### Portfolio - Fees / commissions in position accounting Done when all existing tests pass, each TODO has coverage, and metrics.json is correct end-to-enC++ Trading Engine —d. Stack: C++20, Eigen, CMake, nlohmann/json. Repo is private during the project, open-sourced on completion.

tradingfinanceC++backtestingalgorithmsopen-sourceEigenCMakeOngoingremote
AG
Andrew Garcia
Remote
Apply →
AIfeaturedOpen

Fractal Hunter — Robust Image Classification Under Extreme Distortion

## Up the Gain — Electric & Distortion-Ready Blorbos (toon penguins) hide by transforming into distorted, non-intuitive representations. You are given a dataset derived from a small set of source images, transformed through unknown but structure-preserving processes. 📦 Dataset & full documentation: https://github.com/andrewrgarcia/fractal-hunter-dataset 🔬 Kaggle (for experimentation): https://www.kaggle.com/datasets/drandrewgarcia/fractal-hunter-extreme-distortion ### Task Build a model that determines whether an image originated from Blorbo. ### Constraints - Transformation is unknown - No explicit inversion tricks - Must generalize to unseen images and transformations ### Goals - Robust classification under severe distortion - Clear approach + failure analysis ### Evaluation Held-out test set uses unseen transformations ### Stretch - Learn invariant representations - Recover structure without knowing the transform **Stack:** Python · PyTorch / sklearn

AIML / DataPythonFFTAutoencodersCNN~2 weeksremote
PF
Prince Fractalius
The Fractal Kingdom at Sagittarius A*
Apply →
procedural generationOpen

Make This Skeleton Not Look Like a Snowman

There’s a skeleton example in Moxi. Right now... it’s rough. It looks like a snowman with bones. Your task is to improve it. You can approach this in three ways: 1. Script only: Refine the Moxi script (link below). Improve proportions, structure, and relationships using markdown only. 2. Engine only: Dig into the source code and improve how forms are resolved or rendered. 3. Both: Do whatever it takes to make it better. The goal is a more convincing, more anatomical result. Not perfect realism, but something that actually reads as a skeleton. Start by opening an issue on the Moxi GitHub and outline your approach. Submit: * Your updated script (include your name in headers) AND/OR * Work on a fork on the Moxi Repo Selected contributions will be: * Added to the repo (with credit) * Used to push the system forward Links: * Moxi GitHub: https://github.com/moxilang/moxi * Moxi SKELETON.md script: https://github.com/moxilang/moxi/blob/main/scripts/SKELETON.md

procedural generationcreative codinggamedev3DmarkdownAIopen source~2 weeksremote
AG
Andrew Garcia
Remote
Apply →
Procedural GenerationOpen

Build a World in Markdown (Moxi Script Challenge)

Moxi is a semantic, markdown-based language that generates 3D worlds. This project is simple: write a Moxi script for one of the assets on the Moxi site: [https://andrewcomputing.com/moxi](https://andrewcomputing.com/moxi) * Procedural city block * Modular spaceship The catch: you only work in Moxi code (markdown). No changes to the source code. Submit your script with your name inside the file (use markdown headers). Selected submissions will be: * Featured on the Moxi site * Added to the public repo (with credit) * Used as part of the growing example library This is less about polish and more about ideas. Interesting structure > perfect output. If you want to experiment with a different way of building worlds, this is a good place to start. Rules / Recommendations: * AI is allowed * If you use AI, check the SKILLS.md in the repo Links: * Moxi GitHub: https://github.com/moxilang/moxi * Moxi Site: https://andrewcomputing.com/moxi

Procedural GenerationGamedevMarkdownOpen SourceCreative Coding3DAI~1 weekremote
AG
Andrew Garcia
Remote
Apply →
open-sourceOpen

Citation Extraction Pipeline (Rust, Elaine CLI)

## Extending a Rust CLI for reference management (Elaine) _Deterministic citation extraction from messy PDFs._ GitHub Repo: [elaine-cli](https://github.com/andrewrgarcia/elaine-cli) --- ### The task Improve an existing citation extraction pipeline. It works, but only on some PDFs. Goal: - improve parsing across formats - handle messy layouts - refine deterministic heuristics (no AI) Inspired by [citracer](https://github.com/marcpinet/citracer), but simpler and local-first. Scoped task — not a full rewrite. --- ### What you're working with Prototype exists: - PDF extraction (`pdftotext`) - normalization - reference parsing - CLI (`eln trace`) You’ll get a working branch and build on it. --- ### Success - Works on real PDFs - Fewer parsing failures - Clean, simple code --- ### What you get - Real parsing experience - Work under constraints - Open-source contribution

open-sourceRustgraphsregextext-processingpdfclireference-manager~1 weekremote
AG
Andrew Garcia
Remote
Apply →
Next.jsOpen

Software Engineering Contributor – WorkDog

Work on small, well-defined improvements to the platform. Focus on practical fixes and incremental enhancements. Collaborate on ideas to improve reliability and functionality.

Next.jsTypeScriptFull Stack~2 weeksremote
W
WorkDog
Remote
Apply →
UXOpen

Product & UX Designer – WorkDog

Explore the platform from a user perspective and suggest improvements to usability and clarity. Propose better layouts, flows, and interactions. Work directly with the founder to refine the experience.

UXProductDesign~1 weekremote
W
WorkDog
Remote
Apply →
TestingOpen

Platform Tester – WorkDog

Use the platform as a real user. Test flows like sign-up, project browsing, and applications. Identify bugs, edge cases, and confusing behavior. Share clear and actionable feedback.

TestingQAProduct~1 weekremote
W
WorkDog
Remote
Apply →
GrowthOpen

Growth & Partnerships Associate – WorkDog

Help bring early users to WorkDog by reaching out to student communities, developers, and small networks. Test simple ways to get people to join and participate. Focus on real conversations and practical outreach.

GrowthCommunityOutreach~2 weeksremote
W
WorkDog
Remote
Apply →