Exploring Modern AI Trends and Kubernetes-Based AI Infrastructure
Description
Build a solid understanding of the current landscape of Artificial Intelligence and how modern cloud-native technologies—especially Kubernetes—support AI workloads.
Goals
Use Gemini Learning Mode to guide the exploration, surface relevant concepts, and structure the learning journey:
- Gain insight into the latest AI trends, tools, and architectural concepts.
- Understand how Kubernetes and related cloud-native technologies are used in the AI ecosystem (model training, deployment, orchestration, MLOps).
Resources
Red Hat AI Topic Articles
- https://www.redhat.com/en/topics/ai
Kubeflow Documentation
- https://www.kubeflow.org/docs/
Q4 2025 CNCF Technology Landscape Radar report:
- https://www.cncf.io/announcements/2025/11/11/cncf-and-slashdata-report-finds-leading-ai-tools-gaining-adoption-in-cloud-native-ecosystems/
- https://www.cncf.io/wp-content/uploads/2025/11/cncfreporttechradar_111025a.pdf
Agent-to-Agent (A2A) Protocol
- https://developers.googleblog.com/en/a2a-a-new-era-of-agent-interoperability/
Looking for hackers with the skills:
This project is part of:
Hack Week 25
Activity
Comments
-
about 18 hours ago by jluo | Reply
A great summary from Gemini:
This is a rapidly expanding ecosystem. To keep it organized, I will break this list down by Lifecycle Stage (Training vs. Serving) and Infrastructure Layer (Compute vs. Data).
1. AI Platforms & Orchestration (The "Command Center")
These tools manage the end-to-end lifecycle, gluing everything else together.
- Kubeflow: The "Grandfather" of AI on K8s. It’s a massive suite including:
- Kubeflow Pipelines: For building repeatable workflows (Data -> Train -> Deploy).
- Kubeflow Notebooks: Spawns Jupyter servers as Pods for data scientists.
- Katib: Automated hyperparameter tuning (finding the best learning rate).
- Ray (KubeRay): The top challenger to Kubeflow. It allows you to write Python code that scales across a cluster instantly. It is excellent for both distributed training and serving.
- ZenML: An MLOps framework that sits above the infrastructure, letting you define pipelines in code that can run on Kubeflow, Ray, or simple Kubernetes batches.
2. Training & Scheduling (The "Heavy Lifters")
Standard Kubernetes scheduling (FIFO) is bad for AI training. These tools fix that.
- Volcano: A batch scheduler. It ensures "Gang Scheduling"—meaning if a job needs 50 GPUs but only 49 are available, it waits. (Standard K8s would start 49 and let them sit idle, wasting money).
- Kueue: A newer, lighter alternative to Volcano managed by the K8s specialized interest group. It manages "Job Queues" natively.
- Training Operator: A unified K8s operator that lets you run
PyTorchJob,TFJob(TensorFlow), andXGBoostJobas native K8s objects.
3. Inference & Serving (The "Waiter")
Once a model is trained, these tools serve it to users.
- KServe: The industry standard. It handles "Scale-to-Zero" (via KEDA), canary rollouts, and provides a unified API for TensorFlow, PyTorch, and ONNX models.
- vLLM: The current king of LLM serving. It is highly optimized for GPU memory (PagedAttention) and is often run inside KServe or as a standalone Deployment.
- BentoML / Yatai: A developer-friendly framework. You package your model as a "Bento" (standard format), and Yatai orchestrates the deployment on K8s.
- Seldon Core: An enterprise-grade alternative to KServe with advanced features for compliance, audit trails, and complex inference graphs.
4. Agentic & LLM Ops (The "New Wave")
Tools specifically for the 2025 era of Autonomous Agents.
- LangFlow / Flowise: Low-code "drag-and-drop" UI tools for building LLM chains. They can be deployed on K8s via Helm charts to run agent backends.
- kagent / Agent Sandbox: Emerging tools (often cloud-specific or experimental) that provide secure, isolated environments (using gVisor or microVMs) for agents to execute code safely.
- Ollama: While often used locally, it is increasingly deployed on K8s (via Helm) as a lightweight way to serve open-source models like Llama 3 or Mistral inside a cluster.
5. Data & Memory (The "Brain")
- Vector Databases (with K8s Operators):
- Milvus: A popular open-source vector DB built natively for K8s scalability.
- Weaviate: Another strong option with a solid K8s operator.
- Qdrant: Written in Rust, very fast, and easy to deploy on K8s.
- Feature Stores:
- Feast: The open-source standard for serving features (e.g., "User's last 5 clicks") to models in real-time.
6. Observability & Cost (The "Watchtower")
- Prometheus & Grafana: The standard for metrics (GPU temperature, Request Latency).
- DCGM Exporter: The specific NVIDIA tool that pulls GPU metrics (utilization, memory) so Prometheus can see them.
- KEDA: The autoscaler (discussed previously) that scales pods based on event queues.
- Karpenter: The Node autoscaler. If KEDA asks for more pods, Karpenter instantly buys more EC2/VM nodes from the cloud provider to fit them.
- OpenCost / Kubecost: Tools to track exactly how much money your AI team is spending on GPUs per namespace.
- Kubeflow: The "Grandfather" of AI on K8s. It’s a massive suite including:
-
Similar Projects
MCP Trace Suite by r1chard-lyu
Description
This project plans to create an MCP Trace Suite, a system that consolidates commonly used Linux debugging tools such as bpftrace, perf, and ftrace.
The suite is implemented as an MCP Server. This architecture allows an AI agent to leverage the server to diagnose Linux issues and perform targeted system debugging by remotely executing and retrieving tracing data from these powerful tools.
- Repo: https://github.com/r1chard-lyu/systracesuite
- Demo: Slides
Goals
Build an MCP Server that can integrate various Linux debugging and tracing tools, including bpftrace, perf, ftrace, strace, and others, with support for future expansion of additional tools.
Perform testing by intentionally creating bugs or issues that impact system performance, allowing an AI agent to analyze the root cause and identify the underlying problem.
Resources
- Gemini CLI: https://geminicli.com/
- eBPF: https://ebpf.io/
- bpftrace: https://github.com/bpftrace/bpftrace/
- perf: https://perfwiki.github.io/main/
- ftrace: https://github.com/r1chard-lyu/tracium/
Kubernetes-Based ML Lifecycle Automation by lmiranda
Description
This project aims to build a complete end-to-end Machine Learning pipeline running entirely on Kubernetes, using Go, and containerized ML components.
The pipeline will automate the lifecycle of a machine learning model, including:
- Data ingestion/collection
- Model training as a Kubernetes Job
- Model artifact storage in an S3-compatible registry (e.g. Minio)
- A Go-based deployment controller that automatically deploys new model versions to Kubernetes using Rancher
- A lightweight inference service that loads and serves the latest model
- Monitoring of model performance and service health through Prometheus/Grafana
The outcome is a working prototype of an MLOps workflow that demonstrates how AI workloads can be trained, versioned, deployed, and monitored using the Kubernetes ecosystem.
Goals
By the end of Hack Week, the project should:
Produce a fully functional ML pipeline running on Kubernetes with:
- Data collection job
- Training job container
- Storage and versioning of trained models
- Automated deployment of new model versions
- Model inference API service
- Basic monitoring dashboards
Showcase a Go-based deployment automation component, which scans the model registry and automatically generates & applies Kubernetes manifests for new model versions.
Enable continuous improvement by making the system modular and extensible (e.g., additional models, metrics, autoscaling, or drift detection can be added later).
Prepare a short demo explaining the end-to-end process and how new models flow through the system.
Resources
Multi-agent AI assistant for Linux troubleshooting by doreilly
Description
Explore multi-agent architecture as a way to avoid MCP context rot.
Having one agent with many tools bloats the context with low-level details about tool descriptions, parameter schemas etc which hurts LLM performance. Instead have many specialised agents, each with just the tools it needs for its role. A top level supervisor agent takes the user prompt and delegates to appropriate sub-agents.
Goals
Create an AI assistant with some sub-agents that are specialists at troubleshooting Linux subsystems, e.g. systemd, selinux, firewalld etc. The agents can get information from the system by implementing their own tools with simple function calls, or use tools from MCP servers, e.g. a systemd-agent can use tools from systemd-mcp.
Example prompts/responses:
user$ the system seems slow
assistant$ process foo with pid 12345 is using 1000% cpu ...
user$ I can't connect to the apache webserver
assistant$ the firewall is blocking http ... you can open the port with firewall-cmd --add-port ...
Resources
Language Python. The Python ADK is more mature than Golang.
https://google.github.io/adk-docs/
https://github.com/djoreilly/linux-helper
Song Search with CLAP by gcolangiuli
Description
Contrastive Language-Audio Pretraining (CLAP) is an open-source library that enables the training of a neural network on both Audio and Text descriptions, making it possible to search for Audio using a Text input. Several pre-trained models for song search are already available on huggingface
Goals
Evaluate how CLAP can be used for song searching and determine which types of queries yield the best results by developing a Minimum Viable Product (MVP) in Python. Based on the results of this MVP, future steps could include:
- Music Tagging;
- Free text search;
- Integration with an LLM (for example, with MCP or the OpenAI API) for music suggestions based on your own library.
The code for this project will be entirely written using AI to better explore and demonstrate AI capabilities.
Result
In this MVP we implemented:
- Async Song Analysis with Clap model
- Free Text Search of the songs
- Similar song search based on vector representation
- Containerised version with web interface
We also documented what went well and what can be improved in the use of AI.
You can have a look at the result here:
Future implementation can be related to performance improvement and stability of the analysis.
References
- CLAP: The main model being researched;
- huggingface: Pre-trained models for CLAP;
- Free Music Archive: Creative Commons songs that can be used for testing;
Explore LLM evaluation metrics by thbertoldi
Description
Learn the best practices for evaluating LLM performance with an open-source framework such as DeepEval.
Goals
Curate the knowledge learned during practice and present it to colleagues.
-> Maybe publish a blog post on SUSE's blog?
Resources
https://deepeval.com
https://docs.pactflow.io/docs/bi-directional-contract-testing
Explore LLM evaluation metrics by thbertoldi
Description
Learn the best practices for evaluating LLM performance with an open-source framework such as DeepEval.
Goals
Curate the knowledge learned during practice and present it to colleagues.
-> Maybe publish a blog post on SUSE's blog?
Resources
https://deepeval.com
https://docs.pactflow.io/docs/bi-directional-contract-testing
Bugzilla goes AI - Phase 1 by nwalter
Description
This project, Bugzilla goes AI, aims to boost developer productivity by creating an autonomous AI bug agent during Hackweek. The primary goal is to reduce the time employees spend triaging bugs by integrating Ollama to summarize issues, recommend next steps, and push focused daily reports to a Web Interface.
Goals
To reduce employee time spent on Bugzilla by implementing an AI tool that triages and summarizes bug reports, providing actionable recommendations to the team via Web Interface.
Project Charter
https://docs.google.com/document/d/1HbAvgrg8T3pd1FIx74nEfCObCljpO77zz5In_Jpw4as/edit?usp=sharing## Description
Project Achievements during Hackweek
In this file you can read about what we achieved during Hackweek.
https://docs.google.com/document/d/14gtG9-ZvVpBgkh33Z4AM6iLFWqZcicQPD41MM-Pg0/edit?usp=sharing
SUSE Observability MCP server by drutigliano
Description
The idea is to implement the SUSE Observability Model Context Protocol (MCP) Server as a specialized, middle-tier API designed to translate the complex, high-cardinality observability data from StackState (topology, metrics, and events) into highly structured, contextually rich, and LLM-ready snippets.
This MCP Server abstract the StackState APIs. Its primary function is to serve as a Tool/Function Calling target for AI agents. When an AI receives an alert or a user query (e.g., "What caused the outage?"), the AI calls an MCP Server endpoint. The server then fetches the relevant operational facts, summarizes them, normalizes technical identifiers (like URNs and raw metric names) into natural language concepts, and returns a concise JSON or YAML payload. This payload is then injected directly into the LLM's prompt, ensuring the final diagnosis or action is grounded in real-time, accurate SUSE Observability data, effectively minimizing hallucinations.
Goals
- Grounding AI Responses: Ensure that all AI diagnoses, root cause analyses, and action recommendations are strictly based on verifiable, real-time data retrieved from the SUSE Observability StackState platform.
- Simplifying Data Access: Abstract the complexity of StackState's native APIs (e.g., Time Travel, 4T Data Model) into simple, semantic functions that can be easily invoked by LLM tool-calling mechanisms.
- Data Normalization: Convert complex, technical identifiers (like component URNs, raw metric names, and proprietary health states) into standardized, natural language terms that an LLM can easily reason over.
- Enabling Automated Remediation: Define clear, action-oriented MCP endpoints (e.g., execute_runbook) that allow the AI agent to initiate automated operational workflows (e.g., restarts, scaling) after a diagnosis, closing the loop on observability.
Hackweek STEP
- Create a functional MCP endpoint exposing one (or more) tool(s) to answer queries like "What is the health of service X?") by fetching, normalizing, and returning live StackState data in an LLM-ready format.
Scope
- Implement read-only MCP server that can:
- Connect to a live SUSE Observability instance and authenticate (with API token)
- Use tools to fetch data for a specific component URN (e.g., current health state, metrics, possibly topology neighbors, ...).
- Normalize response fields (e.g., URN to "Service Name," health state DEVIATING to "Unhealthy", raw metrics).
- Return the data as a structured JSON payload compliant with the MCP specification.
Deliverables
- MCP Server v0.1 A running Golang MCP server with at least one tool.
- A README.md and a test script (e.g., curl commands or a simple notebook) showing how an AI agent would call the endpoint and the resulting JSON payload.
Outcome A functional and testable API endpoint that proves the core concept: translating complex StackState data into a simple, LLM-ready format. This provides the foundation for developing AI-driven diagnostics and automated remediation.
Resources
- https://www.honeycomb.io/blog/its-the-end-of-observability-as-we-know-it-and-i-feel-fine
- https://www.datadoghq.com/blog/datadog-remote-mcp-server
- https://modelcontextprotocol.io/specification/2025-06-18/index
- https://modelcontextprotocol.io/docs/develop/build-server
Basic implementation
- https://github.com/drutigliano19/suse-observability-mcp-server
Results
Successfully developed and delivered a fully functional SUSE Observability MCP Server that bridges language models with SUSE Observability's operational data. This project demonstrates how AI agents can perform intelligent troubleshooting and root cause analysis using structured access to real-time infrastructure data.
Example execution
Technical talks at universities by agamez
Description
This project aims to empower the next generation of tech professionals by offering hands-on workshops on containerization and Kubernetes, with a strong focus on open-source technologies. By providing practical experience with these cutting-edge tools and fostering a deep understanding of open-source principles, we aim to bridge the gap between academia and industry.
For now, the scope is limited to Spanish universities, since we already have the contacts and have started some conversations.
Goals
- Technical Skill Development: equip students with the fundamental knowledge and skills to build, deploy, and manage containerized applications using open-source tools like Kubernetes.
- Open-Source Mindset: foster a passion for open-source software, encouraging students to contribute to open-source projects and collaborate with the global developer community.
- Career Readiness: prepare students for industry-relevant roles by exposing them to real-world use cases, best practices, and open-source in companies.
Resources
- Instructors: experienced open-source professionals with deep knowledge of containerization and Kubernetes.
- SUSE Expertise: leverage SUSE's expertise in open-source technologies to provide insights into industry trends and best practices.
Self-Scaling LLM Infrastructure Powered by Rancher by ademicev0
Self-Scaling LLM Infrastructure Powered by Rancher
Description
The Problem
Running LLMs can get expensive and complex pretty quickly.
Today there are typically two choices:
- Use cloud APIs like OpenAI or Anthropic. Easy to start with, but costs add up at scale.
- Self-host everything - set up Kubernetes, figure out GPU scheduling, handle scaling, manage model serving... it's a lot of work.
What if there was a middle ground?
What if infrastructure scaled itself instead of making you scale it?
Can we use existing Rancher capabilities like CAPI, autoscaling, and GitOps to make this simpler instead of building everything from scratch?
Project Repository: github.com/alexander-demicev/llmserverless
What This Project Does
A key feature is hybrid deployment: requests can be routed based on complexity or privacy needs. Simple or low-sensitivity queries can use public APIs (like OpenAI), while complex or private requests are handled in-house on local infrastructure. This flexibility allows balancing cost, privacy, and performance - using cloud for routine tasks and on-premises resources for sensitive or demanding workloads.
A complete, self-scaling LLM infrastructure that:
- Scales to zero when idle (no idle costs)
- Scales up automatically when requests come in
- Adds more nodes when needed, removes them when demand drops
- Runs on any infrastructure - laptop, bare metal, or cloud
Think of it as "serverless for LLMs" - focus on building, the infrastructure handles itself.
How It Works
A combination of open source tools working together:
Flow:
- Users interact with OpenWebUI (chat interface)
- Requests go to LiteLLM Gateway
- LiteLLM routes requests to:
- Ollama (Knative) for local model inference (auto-scales pods)
- Or cloud APIs for fallback
Kubernetes-Based ML Lifecycle Automation by lmiranda
Description
This project aims to build a complete end-to-end Machine Learning pipeline running entirely on Kubernetes, using Go, and containerized ML components.
The pipeline will automate the lifecycle of a machine learning model, including:
- Data ingestion/collection
- Model training as a Kubernetes Job
- Model artifact storage in an S3-compatible registry (e.g. Minio)
- A Go-based deployment controller that automatically deploys new model versions to Kubernetes using Rancher
- A lightweight inference service that loads and serves the latest model
- Monitoring of model performance and service health through Prometheus/Grafana
The outcome is a working prototype of an MLOps workflow that demonstrates how AI workloads can be trained, versioned, deployed, and monitored using the Kubernetes ecosystem.
Goals
By the end of Hack Week, the project should:
Produce a fully functional ML pipeline running on Kubernetes with:
- Data collection job
- Training job container
- Storage and versioning of trained models
- Automated deployment of new model versions
- Model inference API service
- Basic monitoring dashboards
Showcase a Go-based deployment automation component, which scans the model registry and automatically generates & applies Kubernetes manifests for new model versions.
Enable continuous improvement by making the system modular and extensible (e.g., additional models, metrics, autoscaling, or drift detection can be added later).
Prepare a short demo explaining the end-to-end process and how new models flow through the system.
Resources
Cluster API Provider for Harvester by rcase
Project Description
The Cluster API "infrastructure provider" for Harvester, also named CAPHV, makes it possible to use Harvester with Cluster API. This enables people and organisations to create Kubernetes clusters running on VMs created by Harvester using a declarative spec.
The project has been bootstrapped in HackWeek 23, and its code is available here.
Work done in HackWeek 2023
- Have a early working version of the provider available on Rancher Sandbox : *DONE *
- Demonstrated the created cluster can be imported using Rancher Turtles: DONE
- Stretch goal - demonstrate using the new provider with CAPRKE2: DONE and the templates are available on the repo
DONE in HackWeek 24:
- Add more Unit Tests
- Improve Status Conditions for some phases
- Add cloud provider config generation
- Testing with Harvester v1.3.2
- Template improvements
- Issues creation
DONE in 2025 (out of Hackweek)
- Support of ClusterClass
- Add to
clusterctlcommunity providers, you can add it directly withclusterctl - Testing on newer versions of Harvester v1.4.X and v1.5.X
- Support for
clusterctl generate cluster ... - Improve Status Conditions to reflect current state of Infrastructure
- Improve CI (some bugs for release creation)
Goals for HackWeek 2025
- FIRST and FOREMOST, any topic is important to you
- Add e2e testing
- Certify the provider for Rancher Turtles
- Add Machine pool labeling
- Add PCI-e passthrough capabilities.
- Other improvement suggestions are welcome!
Thanks to @isim and Dominic Giebert for their contributions!
Resources
Looking for help from anyone interested in Cluster API (CAPI) or who wants to learn more about Harvester.
This will be an infrastructure provider for Cluster API. Some background reading for the CAPI aspect:
A CLI for Harvester by mohamed.belgaied
Harvester does not officially come with a CLI tool, the user is supposed to interact with Harvester mostly through the UI. Though it is theoretically possible to use kubectl to interact with Harvester, the manipulation of Kubevirt YAML objects is absolutely not user friendly. Inspired by tools like multipass from Canonical to easily and rapidly create one of multiple VMs, I began the development of Harvester CLI. Currently, it works but Harvester CLI needs some love to be up-to-date with Harvester v1.0.2 and needs some bug fixes and improvements as well.
Project Description
Harvester CLI is a command line interface tool written in Go, designed to simplify interfacing with a Harvester cluster as a user. It is especially useful for testing purposes as you can easily and rapidly create VMs in Harvester by providing a simple command such as:
harvester vm create my-vm --count 5
to create 5 VMs named my-vm-01 to my-vm-05.
Harvester CLI is functional but needs a number of improvements: up-to-date functionality with Harvester v1.0.2 (some minor issues right now), modifying the default behaviour to create an opensuse VM instead of an ubuntu VM, solve some bugs, etc.
Github Repo for Harvester CLI: https://github.com/belgaied2/harvester-cli
Done in previous Hackweeks
- Create a Github actions pipeline to automatically integrate Harvester CLI to Homebrew repositories: DONE
- Automatically package Harvester CLI for OpenSUSE / Redhat RPMs or DEBs: DONE
Goal for this Hackweek
The goal for this Hackweek is to bring Harvester CLI up-to-speed with latest Harvester versions (v1.3.X and v1.4.X), and improve the code quality as well as implement some simple features and bug fixes.
Some nice additions might be: * Improve handling of namespaced objects * Add features, such as network management or Load Balancer creation ? * Add more unit tests and, why not, e2e tests * Improve CI * Improve the overall code quality * Test the program and create issues for it
Issue list is here: https://github.com/belgaied2/harvester-cli/issues
Resources
The project is written in Go, and using client-go the Kubernetes Go Client libraries to communicate with the Harvester API (which is Kubernetes in fact).
Welcome contributions are:
- Testing it and creating issues
- Documentation
- Go code improvement
What you might learn
Harvester CLI might be interesting to you if you want to learn more about:
- GitHub Actions
- Harvester as a SUSE Product
- Go programming language
- Kubernetes API
- Kubevirt API objects (Manipulating VMs and VM Configuration in Kubernetes using Kubevirt)
Kubernetes-Based ML Lifecycle Automation by lmiranda
Description
This project aims to build a complete end-to-end Machine Learning pipeline running entirely on Kubernetes, using Go, and containerized ML components.
The pipeline will automate the lifecycle of a machine learning model, including:
- Data ingestion/collection
- Model training as a Kubernetes Job
- Model artifact storage in an S3-compatible registry (e.g. Minio)
- A Go-based deployment controller that automatically deploys new model versions to Kubernetes using Rancher
- A lightweight inference service that loads and serves the latest model
- Monitoring of model performance and service health through Prometheus/Grafana
The outcome is a working prototype of an MLOps workflow that demonstrates how AI workloads can be trained, versioned, deployed, and monitored using the Kubernetes ecosystem.
Goals
By the end of Hack Week, the project should:
Produce a fully functional ML pipeline running on Kubernetes with:
- Data collection job
- Training job container
- Storage and versioning of trained models
- Automated deployment of new model versions
- Model inference API service
- Basic monitoring dashboards
Showcase a Go-based deployment automation component, which scans the model registry and automatically generates & applies Kubernetes manifests for new model versions.
Enable continuous improvement by making the system modular and extensible (e.g., additional models, metrics, autoscaling, or drift detection can be added later).
Prepare a short demo explaining the end-to-end process and how new models flow through the system.
Resources