---
title: "LLM fine-tuning architect: review, role & definition · Undominated.ai"
canonical: https://undominated.ai/agents/wshobson-llm-finetuning-architect/
description: "A fine-tuning planning component that requires a baseline and can recommend a retrieval or prompting alternative."
---

# LLM fine-tuning architect: review, role & definition · Undominated.ai

> A fine-tuning planning component that requires a baseline and can recommend a retrieval or prompting alternative.

[← Explore all agents](/agents/)

MODEL DEVELOPMENT / Seth Hobson

# LLM fine-tuning architect

A fine-tuning planning component that requires a baseline and can recommend a retrieval or prompting alternative.

 [Use this definition ↓](#setup)[Original source ↗](https://github.com/wshobson/agents/blob/46891e7e60da0e52baf1050b7b6391b64e84c6d9/plugins/llm-finetuning/agents/llm-finetuning-architect.md)

SOURCE REVIEW

 Reviewed 2026-10-14
 Evidence 4 linked sources
 Publisher Seth Hobson
 Licence [MIT ↗](https://github.com/wshobson/agents/blob/46891e7e60da0e52baf1050b7b6391b64e84c6d9/LICENSE)
 Revision 46891e7e60da
 [Read what was—and wasn’t—checked ↓](#review)

“Fine-tuning strategist who owns the eval gate and method/model selection. Refuses to plan training without a baselined eval harness. Use PROACTIVELY when a user wants to fine-tune a model, before any training configuration exists.”

 [Seth Hobson · upstream description ↗](https://github.com/wshobson/agents/blob/46891e7e60da0e52baf1050b7b6391b64e84c6d9/plugins/llm-finetuning/agents/llm-finetuning-architect.md) Our analysis follows below.

01 / THE REASONING

## Why this made the selection.

 - It owns planning phases and must hand off a written brief rather than silently train or invent dataset-size guidance.
- The brief includes a replay fraction even when zero is intentionally accepted and cites the source of model and memory assumptions.

### A good fit for

 - Provide the target failure cases, existing evaluation baseline, dataset evidence and compute constraints.
- Produce training-brief.md with method choice, data expectations, replay policy, resource worksheet and promotion criteria.

### Weigh up before choosing

 - No model, dataset rights, hardware or evaluation harness is bundled; a training plan does not establish that fine-tuning helps.
- Source-only component: inspect its upstream context and dependencies before adapting it; no standalone runnable bundle is offered.
- Companion method-selection and quantized-export tables contain fixed data-size, hardware and quality heuristics; they were not benchmark-validated here and must not be treated as universal training or export rules.

02 / THE REVIEW RECORD

## What we actually inspected.

Source review has boundaries. A clear record is more useful than a “safe” badge.

### Material inspected

 - plugins/llm-finetuning/agents/llm-finetuning-architect.md (pinned complete source retained)
- LICENSE
- Repository README and current official host configuration documentation
- plugins/llm-finetuning/skills/finetuning-method-selection/SKILL.md

### Our findings

 - It owns planning phases and must hand off a written brief rather than silently train or invent dataset-size guidance.
- The brief includes a replay fraction even when zero is intentionally accepted and cites the source of model and memory assumptions.

### Not established by this review

 - Source review only: no agent was installed or invoked. Task correctness, effective host restrictions and integration behaviour were not tested.

The review applies to the material and revision named here. A newer upstream release can change its behavior.

03 / PUT IT TO WORK

## Use the role in your project.

[Upstream setup instructions ↗](https://github.com/wshobson/agents/blob/46891e7e60da0e52baf1050b7b6391b64e84c6d9/plugins/llm-finetuning/agents/llm-finetuning-architect.md)
 - Read the pinned source and its licence, then configure an explicitly bounded host context.
- Use the matching fine-tuning skills, current model catalogue and evaluation artifacts; resolve every referenced worksheet and budget before delegating training.
- Provide the target failure cases, existing evaluation baseline, dataset evidence and compute constraints.

### Before you start

 - Matching llm-finetuning plugin reference material
- Dataset rights and evaluation baseline
- Available hardware and explicit compute budget

### Compatibility

Claude Code

### Implementation

Markdown

### Host-mediated agent profile

 - No tools allowlist is declared. Claude Code can inherit or expose the host’s available tools by default; restrict file, shell and network access in the host.

### Cost model

The definition has the stated open-source licence. Model/host usage and any connected service or execution environment can incur separate costs.

04 / FOLLOW THE EVIDENCE

## The source trail.

Our notes are separate from the original resource. Check upstream before adopting a new version.

 - [Pinned upstream agent definition ↗](https://github.com/wshobson/agents/blob/46891e7e60da0e52baf1050b7b6391b64e84c6d9/plugins/llm-finetuning/agents/llm-finetuning-architect.md) Checked 2026-10-14 https://github.com/wshobson/agents/blob/46891e7e60da0e52baf1050b7b6391b64e84c6d9/plugins/llm-finetuning/agents/llm-finetuning-architect.md Supports: Workflow, inputs, outputs and stated boundaries
- [Applicable licence ↗](https://github.com/wshobson/agents/blob/46891e7e60da0e52baf1050b7b6391b64e84c6d9/LICENSE) Checked 2026-10-14 https://github.com/wshobson/agents/blob/46891e7e60da0e52baf1050b7b6391b64e84c6d9/LICENSE Supports: Redistribution and attribution
- [Host configuration and tool inheritance ↗](https://code.claude.com/docs/en/sub-agents.md) Checked 2026-10-14 https://code.claude.com/docs/en/sub-agents.md Supports: Host loading and effective permission boundaries
- [Pinned inspected source: plugins/llm-finetuning/skills/finetuning-method-selection/SKILL.md ↗](https://github.com/wshobson/agents/blob/46891e7e60da0e52baf1050b7b6391b64e84c6d9/plugins/llm-finetuning/skills/finetuning-method-selection/SKILL.md) Checked 2026-10-14 https://github.com/wshobson/agents/blob/46891e7e60da0e52baf1050b7b6391b64e84c6d9/plugins/llm-finetuning/skills/finetuning-method-selection/SKILL.md Supports: Dependency setup and source-review boundary

KEEP COMPARING

## Other approaches to consider.

Related by category or shared topics. These are alternatives to inspect, not a measured quality order.

 [### LLM fine-tuning training engineer ↗ A training component that follows an approved brief and hands checkpoint promotion to a separate evaluator.](/agents/wshobson-llm-finetuning-training-engineer/)[### ML serving and lifecycle engineer ↗ A production-ML planning and implementation profile covering serving, feature validation and model lifecycle checks.](/agents/wshobson-ml-engineer/)[### LLM Fine-Tuning Eval Engineer ↗ Owns the pre-training evaluation baseline and post-training checkpoint verdict, with explicit per-trace records and no authority to train or automatically retrain.](/agents/wshobson-llm-finetuning-eval-engineer/)

 [AI Tools ↗](/tools/)[Skills ↗](/skills/)[Agents ↗](/agents/)[MCP Servers ↗](/mcp-servers/)[Workflows ↗](/workflows/)

## Continue your investigation

 - [Find reusable instructions](/skills/)
- [Inspect connections](/mcp-servers/)
- [Build a toolkit](/tools/)
- [Choose the model](/compare/)
