2stacks/my-resume-v2

Andrew Stanley Resume Q&A

A small, hand-curated chat-format dataset of 122 question/answer pairs covering the professional background, career history, technical skills, certifications, and military service of Andrew Stanley, CTO / Chief Innovation Officer at SMS Data Products Group (McLean, VA).

The dataset is purpose-built for two things:

1. 1.A working demonstration of an end-to-end LLM fine-tuning workflow — source document → synthetic Q&A generation → QLoRA fine-tune → GGUF export → local Ollama deployment.
2. 2.A discoverable, queryable knowledge source about the author — so colleagues, recruiters, and collaborators can ask a fine-tuned local model "who is Andrew?" and get accurate, structured answers.

If you're a hiring manager or peer who landed here: feel free to clone the dataset, fine-tune your favorite base model, and chat with the result. The companion model card (forthcoming) walks through the exact training recipe.

Dataset details

Schema

Each line is a single JSON object with one user turn and one assistant turn:

{"messages": [
  {"role": "user", "content": "Where does Andrew work?"},
  {"role": "assistant", "content": "Andrew works at SMS Data Products Group, headquartered in McLean, VA. ..."}
]}

This format drops in directly to the chat templates used by Qwen3, Llama-3, Mistral-Instruct, and most modern instruction-tuned bases — no preprocessing required for TRL SFTTrainer, Unsloth, or Axolotl.

Topic coverage

• —Identity, current role, location, contact (email only)
• —Career timeline (SMS Data Products Group, prior roles)
• —Sector experience (Defense, Federal, commercial)
• —Technical skills: hybrid multi-cloud (AWS, Azure), DevSecOps, CI/CD, Zero Trust, AI/ML enablement
• —Certifications: AWS AI Practitioner, AWS Solutions Architect Associate, AWS Cloud Practitioner
• —Education: BS in Information Technology / Security, George Mason University (2006)
• —Military service: U.S. Army Signal Officer, Afghanistan deployment 2008–09
• —Security clearance: Active U.S. government clearance (details available upon request)
• —Leadership philosophy and engineering management approach

How this dataset was built

1. 1.Started with Andrew's current resume (PDF) and LinkedIn export as the source of truth.
2. 2.Drafted question/answer pairs by topic cluster — biographical, role-specific, skills, certifications, military, leadership.
3. 3.Wrote each answer to be self-contained (no anaphora across pairs) so the model learns durable facts rather than dialogue continuity.
4. 4.Sanity-checked for PII: phone number scrubbed, only a public-facing email retained.
5. 5.Validated JSONL well-formedness and chat-template compatibility with a Qwen3-8B tokenizer.

Intended use

• —Primary: fine-tune a small/medium open-weight chat model (e.g. Qwen3-8B, Llama-3.1-8B-Instruct, Mistral-7B-Instruct) so it can answer biographical questions about Andrew Stanley.
• —Secondary: reference example for anyone learning QLoRA / instruction tuning who wants a tiny, self-contained dataset to iterate on.

Reference training recipe

The author's reference run used:

A companion model trained with this exact recipe is published at `2stacks/qwen3-8b-andrew-resume-v2` (GGUF Q4KM, drop-in for Ollama / llama.cpp).

Limitations and biases

• —Tiny (122 rows). This is a personal-knowledge fine-tune, not a general capability dataset. Expect heavy overfit to these exact phrasings unless paired with a broader instruction-following dataset during training.
• —Single subject. All assistant outputs concern one real person. Models trained solely on this set will confidently answer about Andrew Stanley and have no grounding for any other entity.
• —Self-curated. The author wrote the answers about himself, so framing reflects how he wants to be described professionally. Treat it as a polished bio, not an independent biography.
• —English-only.
• —Snapshot in time. Reflects career state as of May 2026; will go stale.

Personal information

The dataset intentionally contains:

• —Full name, current employer, city of residence (Bethesda, MD)
• —Public-facing email: 2stacks@2stacks.net
• —Public career history mirroring the author's LinkedIn profile

The dataset intentionally does not contain:

• —Personal phone number
• —Home street address
• —Government / clearance details beyond confirming an active clearance is held
• —Family member information

The subject (Andrew Stanley) is the publisher and consents to release of this information under CC-BY-4.0.

Citation

@dataset{stanley_resume_qa_2026,
  author       = {Stanley, Andrew},
  title        = {Andrew Stanley Resume Q\&A},
  year         = {2026},
  publisher    = {Hugging Face},
  url          = {https://huggingface.co/datasets/2stacks/my-resume-v2}
}

Contact

• —Email: 2stacks@2stacks.net
• —Hugging Face: @2stacks