Projects open for software collabs

We want to develop a Kubernetes Operator that simplifies the deployment of the Coordinator component, currently installed through a Pulumi script and an ArgoCD deployment. The present approach forces us to maintain a fork of the values repository for each customer, which makes installation cumbersome for the end user. The Coordinator consists of twelve microservices written in Go, whose configuration and Helm chart will most likely need to be reworked. The operator will need to handle the installation of the dependencies that are always present, namely Redis and Postfix, as well as the orchestration of those that already ship with their own operator, such as Kafka, Postgres and Prometheus, of which only Postgres and Prometheus are optional since they are sometimes managed directly by the customer. Ideally, we also envision a kubectl plugin to make installation and management as seamless as possible. It is worth noting that Cubbit already relies on other Kubernetes operators to install the many other components of the platform, so this work will fit into an established operational ecosystem.

DittoFS (github.com/marmos91/dittofs, dittofs.io) is a modular virtual filesystem written in Go, capable of serving NFSv3/v4 and SMB2/3 simultaneously in userspace, without relying on FUSE or kernel modules, across interchangeable backends including S3 compatible storage. It integrates deduplication, compression and client side encryption. The project was built over the course of eight months, almost entirely with the support of AI, and it is attracting growing interest from teams looking to run it in production. The product works, but we need someone to read through the code carefully, carry out a technical audit and support us in maintaining the open source repository. In particular, we expect the introduction of reliability tests based on sustained, continuous writes over time, verification of the semantic correctness of permissions between SMB and NFS, the removal of unnecessary code introduced by the AI, and an overall effort to improve performance.

We are also seeing strong interest in building a whitelabel, cross platform version of the Cubbit DS3 Drive application (github.com/marmos91/cubbit-ds3-drive), which was likewise developed with the help of AI. The application is built on a Rust core that, through FFI, generates the bindings for Swift, C# and Kotlin, supporting macOS, iOS, Windows and Android. The macOS and iOS versions are complete and fully functional, while the Windows version is currently in development. On this front we need to complete the Windows version, implement the Android and web versions, and fill a number of product gaps requested by our customers, specifically versioning, proxy support, audit logs, whitelabeling, folder sharing via email, and sharing of large files through an Outlook plugin. Each platform will need to build on the native APIs offered by its respective operating system, such as Apple's FileProvider APIs. Here too, we expect the partner to begin with an audit of the existing code, in order to ensure it is clean, efficient and reliable.