Two server identities. One evolving homelab. A growing collection of self-hosted systems, local AI experiments, software, and technical knowledge.
Primary compute and application machine. Local AI inference, Python workloads, personal assistants, and service processes.
Separate environment for OS testing, CPU-only workloads, network service experiments, and configuration trials.
The existing TAISEN server and process-management system, documented as a distinct component of the homelab.
Personally operated. Independently built. Continuously evolving.
Two server identities with defined roles within the TAISEN ecosystem.
| Attribute | EX19 | SX11 |
|---|---|---|
| Role | Execution | Sandbox |
| Primary focus | AI and compute | Testing and experimentation |
| Processor | i5-9400 | i3-4170 |
| Memory | 16 GB DDR4 | 8 GB DDR3 |
| GPU | GTX 1050 Ti 4 GB | No dedicated GPU recorded |
| Storage | Verify installed drives | 3 TB HDD recorded |
| Typical use | Application workloads and local AI | OS trials and isolated experiments |
Ollama-based model serving and experimentation.
Application backends and personal automation workloads.
Server-management and service processes.
CPU and GPU utilization testing and measurement.
Ubuntu Server compatibility and boot troubleshooting.
Resource measurements without dedicated GPU.
Local service experiments and configuration trials.
BIOS settings, boot media, and storage notes.
Note: The motherboard images above are illustrative hardware references, not verified photographs of the exact boards in these machines. Future servers or proposed roles must be explicitly labeled as planned rather than existing hardware.
Manage the systems that power the lab.
TRFMS is an existing TAISEN server and process-management system. It forms part of the lab's internal management environment and is documented separately from future infrastructure expansion ideas.
Documented health and process visibility.
Existing service-management functions.
Setup, maintenance and troubleshooting.
Relationships with managed applications.
What TRFMS is and why it exists in the homelab.
Only the management and monitoring functions confirmed in the current version.
How TRFMS relates to EX19, SX11, and the services it manages.
Names and descriptions of confirmed managed applications.
Runtime, launch method, dependencies, configuration, and recovery instructions.
Screenshots of the actual TRFMS interface.
Known issues and verified solutions.
Features added, removed, and modified.
Development history note: TRFMS was discussed working locally at 127.0.0.1:8090. This address is not published as a live link — it reflects a development-time configuration, not a visitor-accessible endpoint.
Potential improvements to TRFMS, kept clearly separate from implemented features. Broader infrastructure-management ideas appear here only if still being pursued.
Locally hosted models, inference, embeddings, and personal assistants.
Personal assistant · Local-first memory
KIM is a personal assistant project focused on local-first knowledge management, memory retrieval, and automation.
Personal AI assistant · Local inference
T.U.E.S.D.A.Y. is a personal AI assistant experiment with an emphasis on running language models locally and exploring useful computer-based assistance.
FastAPI application and SQLite database.
Local Ollama model integration.
Embedding models and semantic retrieval.
Persistent knowledge and memory pipeline.
Telegram bot integration.
Local execution and fallback behavior.
Ollama and supported model configurations.
Model size, quantization, and context settings.
Hardware utilization and inference speed.
Assistant capabilities and integrations.
Limits of running models on local hardware.
Planned improvements and experiments.
| Model | Runtime | Host | Task | Quant | Context | Performance | Status |
|---|---|---|---|---|---|---|---|
| sam860/qwen3:1.7b | Ollama | EX19 | Text generation | — | — | To be measured | Experimental |
| Qwen3 embedding workflow | Ollama | EX19 | Embeddings | — | — | To be measured | Experimental |
Models listed as historical experiments. A model is not labeled as currently deployed until its present configuration is verified.
Note: Images above are concept illustrations, not verified screenshots of either application. An optional benchmark archive can record test date, hardware, model configuration, measurements, and methodology.
An index of verified, non-discontinued projects — not a list of every name ever brainstormed.
Server and process-management system forming the lab's internal management environment.
Personal AI and knowledge-retrieval project with local-first memory and automation.
Personal AI assistant project emphasizing local model execution and computer-based assistance.
Execution server for AI, compute, and application workloads. Infrastructure entry, not a software project.
Sandbox and testing server for OS trials, CPU-only workloads, and isolated experiments.
Other programs previously mentioned in the context of TRFMS receive featured project pages only after their current existence and relevance are verified.
What services exist in the homelab, what they do, and where they run.
Services are not claimed to be live merely because their documentation exists. Each entry reflects a last-verified state where confirmation is available.
| Name | Purpose | Host | Runtime | Status |
|---|---|---|---|---|
| TRFMS | Server and process management | To confirm | To confirm | Unverified current state |
| Name | Purpose | Host | Runtime | Status |
|---|---|---|---|---|
| Ollama | Local model serving | EX19 | Native process | Experimental |
| Name | Purpose | Host | Runtime | Status |
|---|---|---|---|---|
| KIM | Knowledge & memory assistant | EX19 | Python / FastAPI | In development |
| T.U.E.S.D.A.Y. | Personal AI assistant | EX19 | Python / Ollama | In development |
Categories such as Automation & Messaging, Development & Testing, and Network & Supporting Infrastructure remain hidden until actual services exist behind them.
How the lab is organized, without exposing details that could compromise it.
Illustrative architecture only. This is not a verified topology or a live network map.
Server connections and physical layout.
Execution workloads vs. sandbox testing.
Concepts, kept sanitized.
Principles, without sensitive detail.
How internal services communicate.
Local and remote connectivity steps.
Privacy: IP addresses, tunnel URLs, router administration pages, credentials, and firewall details are kept private. A more detailed version of the diagram can be retained locally while the published version remains sanitized.
Where TAISEN becomes a long-term technical reference rather than just a portfolio.
Windows configuration, service management, drivers, updates and maintenance.
Verified procedures for operating systems, runtimes, packages and applications.
Symptoms, diagnostic steps, root causes, fixes and verification results.
CPU and GPU utilization, memory consumption, inference performance and latency.
LAN troubleshooting, remote access, name resolution and connectivity diagnostics.
Model comparisons, embedding tests, prompt experiments and inference configurations.
Every published technical note follows a consistent structure, preserving lessons from SX11 OS experiments and EX19 local AI troubleshooting.
| 1 · Date | When the experiment occurred. |
| 2 · Host | EX19, SX11, or another verified machine. |
| 3 · Objective | What you wanted to accomplish. |
| 4 · Environment | Hardware, OS, software versions and relevant settings. |
| 5 · Procedure | What you did. |
| 6 · Result | What happened, including errors. |
| 7 · Resolution | What worked and how you confirmed it. |
| 8 · Lessons learned | What to remember next time. |
| 9 · Status | Resolved, unresolved, superseded, or needs retesting. |
Test results, technical discoveries, and lessons from the lab.
Attempted Ubuntu 26.04 testing on SX11 and encountered a ZSTD corruption error during installation.
Observed GPU and CPU behavior while running local model inference on EX19 during extended standby and active cycles.
TRFMS was confirmed working locally at 127.0.0.1:8090 during an earlier session.
Lab notes represent historical records. Current operation or continued development has not been independently confirmed for all entries.
Distinguishing actual commitments from ideas.
These are proposed website-planning items, not claims of already-committed implementation.
The real TAISEN environment, grouped by subject.
Use real photographs and screenshots for the final site. Redact personal identifiers, tokens, internal addresses, and other sensitive information before publishing.
TAISEN is my personal homelab and technology ecosystem, built around self-hosted infrastructure, software development, local artificial intelligence, automation, and continuous experimentation.
It brings together the machines I operate, the applications I develop, the systems I maintain, and the technical knowledge I accumulate along the way.
EX19 serves as the execution and AI/compute machine, while SX11 provides a separate environment for sandboxing, testing, and experimentation. Each has a defined role within the broader lab.
TRFMS is part of the existing server and process-management environment. Alongside it, my software and AI projects explore practical ways to use local computing resources, manage information, and build useful tools.
TAISEN is built around learning by doing, understanding the systems I operate, maintaining control over my infrastructure, and documenting both successful experiments and failures.
Build a homelab that is understandable, maintainable, useful, and capable of evolving over time, while keeping its infrastructure and technical knowledge organized in one place.
Publication rule: Not every experiment becomes a finished product. Not every planned feature is implemented. Each project is documented according to what actually exists and what has been verified.