AI Engineering · Philadelphia, PA

Learn it. Build it.
Govern it. Prove it.

Practical AI engineering learning resources and editable planning templates for documentation and proposal work. Each resource is designed to show you real inputs, actual outputs, and honest limits — so you know what you're getting before you buy.

LLM Engineering Defense Tech Practical AI Real Business Results
Explore the products

Six resources. Honest descriptions.

Know what you're getting

Learning materials help you build skills. Templates help organize your work. Neither replaces implementation, evidence, or qualified review. Choose the resource that fits the task you actually need to complete.

AI Engineering Workbook

Learn the model-development workflow

For Python developers learning how to prepare data, train or adapt a model, evaluate results, and plan deployment decisions.

An educational workbook with code examples and project checklists. The fourteen-notebook collection and the PDF edition use different chapter sequences — follow the map for the edition in your download.

Python familiarity required. Examples need testing in your own environment. Start with the reviewed CPU notebooks before GPU experiments.
$49 one-time

Interactive AI Walkthroughs

Explore AI concepts in your browser

For learners who want to see how a setting or engineering decision changes a visual demonstration before modifying Python code.

Seventeen HTML walkthroughs with a topic index. These are browser teaching demonstrations — not seventeen Colab notebooks. The separately available Python collection contains fourteen .ipynb notebooks with a different sequence.

Local controls illustrate concepts; they do not train or query a live language model. Complete browser interaction testing is ongoing.
$69 one-time

CMMC Level 2 Readiness Templates

Organize your system documentation and evidence

For organizations documenting systems and safeguards covered by their contractual security obligations.

Editable SSP, boundary worksheet, and security-policy development documents; plus asset, data-flow, evidence, and action registers; and a 110-identifier Revision 2 requirement tracker.

Templates do not implement safeguards, establish compliance, calculate an official SPRS score, or guarantee certification. Verify your applicable baseline and obtain qualified review before making any assessment representations.
$349 one-time

SBIR Phase I Development Template

Turn a research idea into a structured proposal plan

For small-business applicants who need to connect a technical problem, measurable objectives, research tasks, and deliverables into a structured draft.

An editable six-page development template, solicitation compliance matrix, and work-plan and budget-assumption register. Start with the current official topic instructions before drafting.

This is a development framework, not a completed or submitted proposal. Current solicitation requirements, official forms, and authorized submission still apply. Not a guarantee of eligibility or award.
$149 one-time

AI Developer's Reference

A technical reference across the model-development stack

For technical readers who want a structured reference covering problem definition, datasets, preprocessing, architecture, fine-tuning, evaluation, deployment, RLHF, and RAG.

The revised ten-chapter workbook with two appendices. This edition overlaps with the earlier eight-chapter workbook — select the edition whose depth and sequence fits your project stage.

A reference book is not a turnkey application. Code examples in PDF are not freshly executed; test them in your own environment.
$99 one-time

JAWNVION AI Learning Bundle

One coordinated learning path

For technical learners who want the workbook and visual walkthroughs together with a shared topic map.

The selected workbook edition, seventeen browser walkthroughs, and the companion notebook collection when included in the verified download manifest. Read one topic, explore its walkthrough, then run the matching notebook.

Confirm the exact package contents before purchase. GPU availability and training setup time vary. Start with the reviewed CPU lessons.
$99 one-time

EKOS — Knowledge Authority Engine

Search what you know.
Flag what you don't.

EKOS is not a chatbot and not a RAG system. It is an authority engine — every claim is tagged as Known, Inferred, or Unknown. It never fabricates across a gap it can't fill. The verified local demonstration is available now; enterprise features are on the roadmap.

Known

Verified, sourced, authoritative. Confidence ≥ 0.70.

Inferred

Derived with tracked confidence. Score 0.30–0.70.

Unknown

Explicit gap marker. Answer text is null. No fabrication.

The verified local demonstration imports supplied records and returns traceable excerpts using keyword search. It cannot recover conversations absent from your export or prove that a source is true. Semantic retrieval, CUI security enclave, contradiction detection, and enterprise deployment are development roadmap items — not part of the current verified release.

Request a demonstration Enterprise inquiry

Before you choose

Do I need Python experience?

The code notebooks are for technical learners who can follow Python. The seventeen HTML walkthroughs illustrate concepts in a browser without writing code. The CMMC and SBIR templates are editable working files that do not require programming.

Are the CMMC templates a certification or official score?

No. The templates organize documentation and track requirement identifiers. Your organization must implement applicable safeguards, document its actual environment, and provide evidence through the assessment path required by your contract. A filled-out document is not implemented security.

Will the SBIR template guarantee a Phase I award?

No. A strong application requires a suitable current topic, credible and measurable research claims, compliance with the latest solicitation instructions, and authorized submission through the official portal. Funding decisions are outside the template's control.

What is the difference between the workbook, the labs, and the playbook?

The eight-chapter workbook and fourteen-notebook collection provide a structured hands-on learning path with runnable code. The seventeen HTML walkthroughs are browser demonstrations — not Colab notebooks. The ten-chapter developer reference is a more detailed engineering reference that overlaps with the workbook. Review each description and confirm what is in your download before purchasing.

What should I verify before purchase?

Ask for the exact edition included, the file manifest, current test notes, and setup requirements. Confirm that the download matches what the product description says. For bundles, verify which notebook collection is included.

Can I use the training experiments commercially?

The GPU training demonstrations used the Alpaca dataset, which carries a CC BY-NC 4.0 non-commercial license. Saved training reports are teaching evidence. Any weights produced in those runs are research demonstrations and are not distributed as commercially cleared products.

Built around practical work

JAWNVION LLC is led by Peter Terefenko, an AI engineer and defense tech founder based in Philadelphia. The focus is making technical AI work easier to understand, document, and prove — through clear demonstrations, honest product descriptions, and rigorous knowledge engineering.

EKOS is JAWNVION's core technology: an enterprise knowledge authority engine that tracks provenance, flags gaps, and never fabricates across missing evidence. CAGE Code: 24TQ0.

6 Live products
14 Notebooks
17 HTML labs
110 CMMC controls

Questions about a product, your download, or enterprise EKOS access? Email directly:

peter@jawnvion.com

Live program

AI for Real Business

A four-week hands-on cohort for business owners and operators who want to understand, evaluate, and apply AI — without the hype. Define a real problem, build a working solution, and leave with a governance framework you can defend.

Reserve your seat — $897