Bertil - Edge AI Engineer AI Skill
$7.00
Sale price
$7.00
Regular price
Skip to product information
Get Unlimited · $59/yr
Bertil - Edge AI Engineer AI Skill
$7.00
this skill
one payment
yours forever
// skill previewbertil-model-distillation-on-device.md
# Bertil - Model Distillation & On-Device Engineer
## Who Bertil Is
Drop Bertil into Claude and get a Model Distillation and On-Device Engineer who treats the
phone in the technician's hand as the real deployment target and every number as something
that must be measured on that phone rather than inferred from a leaderboard. Bertil starts
from three assumptions that survive contact with reality: a general benchmark tells you
almost nothing about whether a 4-bit model still extracts your eleven work-order fields, a
tokens-per-second figure from the first thirty seconds of a run tells you nothing about
minute six, and "it fits in RAM" is not the same claim as "the operating system will let
your app keep it there". He measures the student against the teacher on the client's own
task, at the client's own quantization, on the client's own worst device.
Bertil covers the whole path from a large model that works to a small model that ships:
teacher output generation and rationale supervision, response, logit and hidden-state
distillation, task-specific student selection and sizing, pruning and sparsity where it
Preview: lines 27 to 34 of 239 · Full file after purchase
or included in Unlimited Access
Use the name
Once the file is loaded, talk to it by name: “Serge, check this page.” That is what the name is for. It also keeps several skills apart in one chat.
Once the file is loaded, talk to it by name: “Serge, check this page.” That is what the name is for. It also keeps several skills apart in one chat.
// the skill your AI runs on
You are Bertil, a Model Distillation and On-Device Engineer who measures a student against its teacher on the client's own task, at the client's own quantization, on their worst device. You have been activated to ship a small model.
$4.92/mo, billed yearly. Works in Claude, ChatGPT, Claude Code, Codex and Cursor. Cancel anytime, 14-day refund on the first charge. After checkout we email your setup link.
Complete skill package instant downloadbertil-model-distillation-on-device.md
Pay once, keep foreverInstant download30-day money-back guarantee
Secure checkout by Shopify
Trademarks of their respective owners. KissMySkills is not affiliated with or endorsed by them.
Ask Bertil something hard.
We need this running on the phone, offline. Shrink the model without wrecking it.
CLAUDE · AS BERTIL, EDGE AI ENGINEER
Bertil returns a distillation plan with the teacher and the target task set, a quantization comparison with quality measured at each level rather than assumed, a runtime selection for the actual target hardware, measured latency, memory, thermal and battery figures on device, a hybrid design stating which requests escalate to the large model, and an honest statement of the tasks where the small model is not good enough.
Or have every skill inside your AI.
Questions before you buy? A person answers, usually the same day: hello@kissmyskills.com
// what's inside
What's inside this skill
- Response, feature and rationale distillation from a teacher
- INT8/INT4, GPTQ, AWQ and GGUF with measured quality cost
- llama.cpp, MLX, Core ML, ONNX and TFLite on real hardware
- Memory, thermal and battery budgets; hybrid escalation patterns
Teams that need a model running offline or on-device without paying for it in accuracy they did not measure.
What you're actually buying
Drop Bertil into Claude and get a distillation engineer who measures the quality cost of every quantization step instead of assuming INT4 is free.
Bertil makes a small model good enough and gets it onto the device: knowledge distillation from a large teacher across response, feature and rationale distillation; task-specific small language models; quantization across INT8, INT4, GPTQ, AWQ and GGUF with the quality cost measured rather than assumed; pruning and sparsity; LoRA and adapter merging; on-device runtimes including llama.cpp, MLX, ONNX Runtime, Core ML and TFLite; memory, thermal and battery budgets on real hardware; latency on mobile and edge silicon; the hybrid pattern where a small model handles the common path and escalates the rest; offline capability and update distribution; and honest evaluation against the teacher on the tasks that matter.
What you get
- →Response, feature and rationale distillation from a teacher
- →INT8/INT4, GPTQ, AWQ and GGUF with measured quality cost
- →llama.cpp, MLX, Core ML, ONNX and TFLite on real hardware
- →Memory, thermal and battery budgets; hybrid escalation patterns
📄 bertil-model-distillation-on-device.skill
Under 2 min install
Works with Claude, ChatGPT & any AI chat
How to install
Download the .skill package → open Claude → paste SKILL.md into your Project Instructions or system prompt → describe your requirement → Bertil builds the answer. Includes a full worked example so you see exactly what you get.
// how to install
Under 2 minutes
Four steps. Any AI chat.
- 01Download the file
After checkout, the download link lands in your inbox. Save the file anywhere on your device.
- 02Open your AI chat
Claude, ChatGPT, Gemini, Grok, or Copilot - whichever one you already use.
- 03Paste the file contents
Drop it into the system prompt, Project instructions, or custom instructions field.
- 04Start working
Your AI is now configured as a specialist. Ask it anything inside its domain.
No technical knowledge required.
With Unlimited there is nothing to download. Connect your AI once, then just ask: "Load Bertil from KissMySkills."
// faq
Questions about this product
What does the Bertil skill do?+
Get a small model shipping: distill it, quantize it with the quality cost measured, and hold the device budget. Load it once into Claude Projects and you get a configured Model Distillation & On-Device Engineer without re-explaining context at the start of every session. Measure the model on your own data before trusting any public benchmark, keep a human in the loop wherever a wrong answer is expensive, and state the failure modes as plainly as the wins.
How do I install this skill file?+
Download the .skill package (it contains SKILL.md), paste the contents into Claude Projects Instructions or your AI's system prompt, add your own context and start your first session. Works with Claude, ChatGPT, or any AI chat that accepts system prompts.
Which AI tools does this skill work with?+
Works with Claude (recommended), ChatGPT, Gemini, Perplexity and Copilot, and any AI chat that accepts system prompts. Claude Projects gives the best results.
What is included in this download?+
One .skill package delivered instantly after purchase: the full SKILL.md role configuration plus a worked-example file with a real scenario so you see the quality before you rely on it. Pay once, keep forever, yours permanently.
How is this different from using Claude without the Bertil skill?+
Without a skill file your AI starts every session as a general assistant. With Bertil loaded it applies Model Distillation & On-Device Engineer methodology from the first message, with consistent quality every time. Measure the model on your own data before trusting any public benchmark, keep a human in the loop wherever a wrong answer is expensive, and state the failure modes as plainly as the wins.
Do I have to read it myself?+
No. The file is written for your AI to read, not for you. Upload it or paste it in once and the AI takes on the role. After that you just ask it questions the way you normally would. You're welcome to open it and read it, but nothing here depends on you doing that.