AI & Intelligence
The core of the lab: how large models reason, retrieve, call tools and run on hardware you control.
Independent AI Lab
Exploring Intelligence.
Building What's Next.
Learning, experimenting and creating with artificial intelligence.
探索人工智能，创造智能产品，让技术真正落地。
Introduction
A Space for Exploration.
LogeeAI is where curiosity meets technology.
A personal laboratory dedicated to exploring artificial intelligence, building intelligent applications, and experimenting with emerging technologies.
- 01LearnBuilding a working knowledge of how intelligent systems actually behave.
- 02ExperimentTesting models, ideas and tools before believing anything about them.
- 03CreateTurning what works into software people can genuinely use.
- 04DeployTaking it the last mile — servers, devices, networks and on-device inference.
Research Areas
Explore the Possibilities.
Four directions. One continuous journey.
Creative Development
Turning model capability into products — interfaces, integrations and small tools that earn their place.
Edge & IoT
Intelligence that lives on the device: sensing, messaging and inference far from a data centre.
Engineering & Deployment
The unglamorous half of the work — getting models and services to run reliably, and stay running.
Featured Projects
Ideas into Reality.
Selected creations and ongoing explorations.
View allAI Assistant
A private conversational workspace for research and writing — retrieval over your own notes, tool access, and memory that stays on infrastructure you own.
View projectIn development
Local Intelligence
Running capable open-weight models on hardware you own: quantisation, inference tuning and privacy-first workflows.
View projectIn development
Edge AI System
A distributed sensing stack where small models make decisions on the device instead of waiting on the cloud.
View projectIn development
All entries above are concepts in development. Nothing on this page has been released yet.
Experiments
Inside the Lab.
A collection of experiments, discoveries and technical explorations.
View allLocal LLM Exploration
Measuring inference speed and memory footprint of open-weight models on consumer hardware, and where quality actually degrades.
AI Agent Experiments
Tool-calling loops, planning strategies and the failure modes that show up once an agent runs for more than a few steps.
IoT & Edge Computing
MQTT pipelines and on-device inference across ESP32 and Raspberry Pi nodes, from sensor to decision without a round trip.
Lab log entries are logged as they happen. Dates and results stay empty until there is something real to report.
Journal
Notes from the Lab.
Working notes on models, tooling and the things that broke along the way.
View allFoundations01
Getting Started with Large Language Models
Tokens, context windows and sampling — the parts of a language model that determine what it can and cannot do.
DraftReading time TBD
Agents02
Building Your First AI Agent
From a single tool call to a loop that plans, acts and checks its own work without running away with your API budget.
DraftReading time TBD
Engineering03
Deploying a Local AI Model
Serving a model on your own machine: runtimes, quantisation, memory limits and the trade-offs nobody mentions.
DraftReading time TBD
These titles are sample entries that show the shape of the journal. No article has been published yet.
About
Driven by Curiosity.
- 01Build to understand, not to demonstrate.
- 02Prefer a working prototype over a convincing slide.
- 03Keep systems small enough to reason about.
LogeeAI is an independent space for learning, experimentation and creation.
I believe the best way to understand technology is to build something with it.
This site is the public notebook of that process — the questions worth chasing, the experiments that fail, and the products that come out the other side.
Everything here is built independently: research, design, engineering and deployment.
LogeeAI.Think. Build. Explore.