You can fine-tune a small custom LLM on Ubuntu by running Charmed Kubeflow on MicroK8s. Feast holds the reusable features, a Kubeflow Trainer v2 job runs a Hugging Face LoRA fine-tune, a Kubernetes volume stores the checkpoint, and KServe serves it. Canonical’s Rob Gibbon lays out this route in a tutorial dated September 23, 2026. This article explains how that tutorial fits together, what it needs, and where it stops being a recipe you can copy blindly.
Treat the tutorial as a worked example, not a tested universal recipe. Its commands pin specific versions and a release-candidate branch, so check compatibility before you reuse them.
As an Amazon Associate I earn from qualifying purchases.
What the walkthrough builds
The tutorial chains five stages, each handled by a different component. The model is not trained by Feast. Feast supplies and organizes the data the trainer consumes.
| Stage | Component | What happens |
|---|---|---|
| 1. Platform | Charmed Kubeflow on MicroK8s | A compact cluster with Feast, Kubeflow Trainer v2 and KServe enabled |
| 2. Data | Hugging Face dataset, PostgreSQL | The nampdn-ai/tiny-webtext dataset is downloaded and ingested into PostgreSQL |
| 3. Features | Feast | Features are registered in a Feast store backed by PostgreSQL, so they are reusable |
| 4. Training | Kubeflow Trainer v2, Hugging Face LoRA | A training job reads the Feast features and fine-tunes the model with LoRA |
| 5. Serving | PersistentVolumeClaim, KServe | The checkpoint is saved on a PVC, and a KServe InferenceService loads it from there |
All of this comes from the Ubuntu tutorial. The files it supplies are feature definitions, ingestion and training scripts, a Trainer v2 job manifest, a distributed-training script variant, dependency declarations and a PVC manifest. The distributed variant is there for scaling out. The main walkthrough still uses a single training instance.
#1 Best Overall
- UNOPENED RETAIL PACKAGING, sold as configured by Lenovo. Includes one year of Courier or Carry-in Lenovo Warranty. Add up to 5 years of Lenovo Premier Onsite Support Plus when you register your computer with Lenovo.
- The ThinkPad P16s Gen 4 is a compact mobile workstation powered by an AMD Ryzen AI 7 PRO 350 processor, offering premium AI performance and real-time workload optimization. It also features a numeric keypad to boost productivity and an extended battery life for all-day power.
- With 32 GB DDR5-5600MT memory and a 1 TB SSD, the Copilot+ mobile workstation's dedicated AI-driven neural processing unit enhances productivity by automating tasks, optimizing workflows, and delivering top-tier performance.
- Plenty of connectivity: 1x USB-A (USB 5Gbps / USB 3.2 Gen 1); 1x USB-A (USB 5Gbps / USB 3.2 Gen 1), Always On; 2x USB-C (Thunderbolt 4 / USB4 40Gbps), with PD 3.0 and DisplayPort 1.4; 1x HDMI 2.1, up to 4K/60Hz; 1x Headphone / microphone combo jack (3.5mm); 1x Ethernet (RJ-45); and 1x Security keyhole.
- The mobile workstation is a visual splendor, whether editing designs or creating content, the OLED touchscreen display is excellent for any project. Equipped with high speed WiFi 7 and a 5MP RGB+IR camera with premium mics.
What you need before starting
Host and account requirements
The fine-tuning tutorial asks for one or more computers running Ubuntu 24.04 LTS or later, with 32 GB of RAM and 16 CPU cores. It says more is better. You also need a stable internet connection, basic Linux and Kubernetes familiarity, and a Hugging Face account. The walkthrough does not state a GPU requirement for its one-worker example. It presents GPU accelerators and multi-node infrastructure as ways to scale up.
Two guides, two baselines
Canonical’s separate Charmed Feast getting-started tutorial lists different requirements. Don’t merge the two.
Rank #2
- Unopened retail packaging, sold as configured by Lenovo. One Year Courier or Carry In Lenovo Warranty. Add up to 5 years of coverage when you register your computer with Lenovo.
- The 14” Lenovo ThinkPad P14s Gen 6, Lenovo’s thinnest and lightest mobile workstation, boasts unmatched power with the AMD Ryzen AI 7 PRO 350 processor, delivering supreme AI performance for real-time workload optimization. This Copilot+ PC features AMD Radeon integrated graphics for intensive AI workflows for amplified productivity and efficiency.
- This mobile workstation is designed for business professionals, offering powerful performance with its advanced processor and ample memory, ensuring smooth multitasking and efficient workflows. The vibrant 14" display with high brightness and color accuracy is perfect for detailed work, while the long-lasting battery supports productivity on the go. While ideal for professionals, its robust features make it a great choice for anyone seeking a reliable and high-performing laptop.
- Plenty of ports, including: 1x USB-A (USB 5Gbps / USB 3.2 Gen 1); 1x USB-A (USB 5Gbps / USB 3.2 Gen 1), Always On; 2x USB-C (Thunderbolt 4 / USB4 40Gbps), with PD 3.0 and DisplayPort 1.4; 1x HDMI 2.1, up to 4K/60Hz; 1x Headphone / microphone combo jack (3.5mm); 1x Ethernet (RJ-45); and 1x Security keyhole.
- Boost your productivity with the Copilot+ mobile workstation. With a dedicated AI-driven neural processing unit, it revolutionizes work by crunching datasets, automating repetitive tasks, and optimizing workflows. Enjoy top-tier performance paired with exceptional efficiency for the most demanding tasks.
| Guide | OS | CPU | RAM | Disk |
|---|---|---|---|---|
| Fine-tuning walkthrough (Ubuntu, September 2026) | Ubuntu 24.04 LTS or later | 16 cores | 32 GB | not stated |
| Charmed Feast getting started (Canonical, publication date not stated) | Ubuntu 22.04 or later | 4 cores | 32 GB | 50 GB available |
If you are sizing a machine for the full LLM workflow, the fine-tuning guide’s figures are the relevant ones. Neither guide endorses a hardware brand or model. A workstation matching “32 GB RAM, 16 cores” meets the stated baseline.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
Pinned versions
The tutorial’s commands use these choices:
- MicroK8s channel
1.34-strict/stable - Juju channel
3.6/stable - The
track/1.11-rcbranch of thecharmed-kubeflow-solutionsrepository - A Terraform configuration that enables Feast, KServe and Training v2 and disables several other modules
These are the author’s choices as of September 23, 2026. They are not compatibility guarantees. The branch name ends in -rc, which suggests a release candidate. Before copying commands, check that the channels and branch still line up with the Charmed Kubeflow release you intend to run.
Rank #3
- DESIGNED FOR PROFESSIONALS ON THE MOVE - The Dell Precision 3490 marries professional-grade performance with portability to elevate your work-anywhere experience. Weighing just 3.09 lbs and tested to MIL-STD 810H military standards, it hits the sweet balance: delivering the robustness and power for demanding applications, sans the flagship Precision 5690’s premium price or the desktop-replacement Precision 7680’s excessive heft. Enjoy seamless productivity on this single, powerful workstation.
- PREMIUM PERFORMANCE - Powered by the Intel Core Ultra 5 135H Processor (14 Cores, up to 4.6GHz) and Intel graphics, this laptop delivers seamless multitasking and creativity, plus AI-assisted productivity to boost workflow efficiency. It also features 32GB DDR5 RAM and 1TB SSD for fast storage and reduced load times, ensuring smooth and responsive performance for all your tasks.
- CRISP DISPLAY & PRIVACY - 14" FHD (1920×1080) display delivers vibrant and comfortable viewing for everyday professional work. Support for up to 3 external monitors via HDMI and Thunderbolt ports at 4K@60Hz (without docking station). A built‑in 1080p FHD HDR RGB webcam with privacy shutter ensures clear, reliable video calls for collaboration and meetings.
- VERSATILE CONNECTIVITY - Equipped with two Thunderbolt 4, two USB-A, HDMI, Ethernet, and an Audio combo jack for flexible connections. With Wi-Fi 6 and Bluetooth, ensuring fast wireless connectivity and compatibility with a wide range of peripherals. Working comfortably in any lighting with a backlit keyboard.
- OPERATING SYSTEM - Windows 11 Pro 64‑bit, with AI‑powered Copilot, offers intelligent assistance to streamline complex professional workflows, enhance productivity, and support advanced multitasking across demanding applications. Built for workstation‑class computing, it delivers enterprise‑grade security and IT manageability.
How Feast fits in
Feast is an open-source feature store for defining, managing, validating and serving model features. Kubeflow’s Feast introduction describes why that matters. Point-in-time-correct retrieval and consistency between training and serving help avoid the mismatched features that can contribute to performance drops in production. The same page lists an offline store, an online store, a registry and a workflow engine among the general integration requirements.
In this walkthrough, PostgreSQL backs the Feast store, the features are registered once, and the training job reads from them. For text fine-tuning that is a modest use of Feast. The benefit is reuse: later experiments can pull the same registered definitions instead of re-implementing preprocessing in each script.
Rank #4
- Intel Core i9-13950HX Processor for demanding professional applications and multitasking workloads. Includes Dell Manufacturer Warranty through March 2031.
- Professional Workstation Configuration – Designed for engineering, design, software development, data analysis, and other business applications.
- NVIDIA RTX 3500 Ada Generation: Featuring 12GB of VRAM, this professional-grade GPU delivers the stability and power required for advanced engineering, architectural design, and intensive content creation.
- Built for Business & Connectivity – Features HDMI, USB-C, Wi-Fi, Bluetooth, and Windows 11 Pro with AI Copilot for productivity, security, and modern workflows.
- ISV-Certified Workstation Performance – Optimized and tested for professional software applications used in design, engineering, and data science.
On the platform side, Charmed Kubeflow’s architecture documentation says its Resource Dispatcher gives user namespaces Feast credentials and a feature_store.yaml. Users can then run Feast commands from Notebook servers. Your deployment still needs its database services and credentials configured.
Free tools Windows power users keep installed
One-click scans. No signup required.
Walking through the stages
1. Deploy the platform
You install MicroK8s and Juju at the pinned channels. You then apply the Terraform configuration from the charmed-kubeflow-solutions branch. It brings up Charmed Kubeflow with only the modules this workflow needs: Feast, KServe and Training v2. Keeping the module set small is what makes the one-node footprint workable.
Best Value
- [AI-OPTIMIZED POWER IN A COMPACT BUILD] The 14” Lenovo ThinkPad P14s Gen 6, a thin and light mobile workstation, boasts unmatched power with AMD Ryzen AI PRO 300 Series processors, delivering supreme AI performance for real-time workload optimization. This Copilot+ PC features AMD Radeon integrated graphics for intensive AI workflows for amplified productivity and efficiency. Features Zen 5 Gen Ryzen AI 7 350 2.00GHz Processor (upto 5 GHz, 16MB Cache, 8-Cores, 16-Threads) and AMD Radeon 860M Integrated Graphics
- [CLEAR AND COMFORTABLE VIEWING ALL DAY] Features 14.0" IPS WUXGA (1920x1200) 60Hz Display; 65W PSU, Type-C Power-In, 4-Cell 57 WHr Battery; Black Color
- [HIGH-SPEED COLLABORATION WITHOUT THE HASSLE] Stay ahead and connected with advanced WiFi with seamless speed. Designed with a robust port selection and lightning-fast memory, this device ensures you enjoy seamless, high-speed collaboration and rapid data transfers, making it perfect for juggling demanding tasks. Tailored for power users, it delivers reliable performance without any compromises. Features 16GB DDR5 SODIMM, 512GB PCIe NVMe SSD; 802.11be, Bluetooth 5.4, RJ-45, Webcam, 1 x HDMI 2.1, 2 Thunderbolt 4, Headphone/Microphone Combo Jack.
- [PROFESSIONAL-GRADE OPERATING SYSTEM] Windows 11 Pro 64-bit provides advanced security tools, business-class management features, and AI-powered Copilot to simplify everyday tasks. Ideal for professionals, educators, creators, remote workers, and anyone needing a dependable platform for virtual meetings, streaming, and multitasking.
- [PROFESSIONAL UPGRADE] The original seal has been opened only to perform authorized hardware upgrades. The upgraded RAM/SSD is covered by a 3-year warranty from MichaelElectronics2, while all remaining components continue under the original 1-year manufacturer warranty.
2. Ingest data and register features
The ingestion script pulls nampdn-ai/tiny-webtext from Hugging Face into PostgreSQL. The Feast feature definitions then describe that data as registered features. This is the step where your Hugging Face account matters.
3. Run the Trainer v2 job
The Trainer v2 job manifest launches the training script. The script retrieves features from Feast and fine-tunes with Hugging Face LoRA. LoRA trains small adapter weights rather than updating every parameter, which is why a compact setup is plausible for a small model. The author’s rough estimate is that training may take an hour or more depending on your hardware. It is not a performance promise.
4. Persist the checkpoint
The tutorial’s PVC manifest requests 20 GB, and the trained checkpoint is written there. That size is an example choice, not proof that every model checkpoint fits. If you swap in a larger base model, size the claim accordingly.
5. Serve with KServe
You create a KServe InferenceService using a Hugging Face predictor whose storage URI points at the PVC. KServe loads the checkpoint from the volume and exposes an OpenAI-style chat completions route. The tutorial shows a sample chatbot reply. That reply only demonstrates how to call the endpoint. It says nothing reliable about model quality.
Limits to keep in mind
- It is not a capacity benchmark. The tutorial demonstrates a compact, single-worker workflow. It says larger deployments can use many nodes, GPU accelerators and high-performance networking, but it gives no sizing rule for them.
- Kubeflow’s LLM hyperparameter tuning is optional and alpha. The Katib LLM hyperparameter optimization guide labels the feature alpha. It lists
train_lossas the currently supported LLM objective metric. It also says distributed training is not supported for the described custom-objective path. The core tutorial does not need it. - Your Feast setup may differ. Feast’s own Feast on Kubernetes guide gives a general deployment overview. The tutorial’s PostgreSQL-backed arrangement is one configuration among several.
Self-managed or managed
The tutorial itself is self-managed: you run a one-node cluster and operate it yourself. It also mentions a managed Charmed Kubeflow option that runs in the customer’s Microsoft Azure tenancy, with operations handled by Canonical. The tutorial gives no pricing or service-level terms for that option, so confirm the current offer with Canonical before choosing it. The practical split comes down to who operates the platform, where it runs and how far you expect to scale. The one-node setup suits learning and prototyping. Anything larger raises the operations and infrastructure questions the tutorial leaves unanswered.
