LLM & Reasoning
Exploring large language models, prompt engineering, reasoning, and evaluation.
- Large Language Models
- Prompt Engineering
- Model Reasoning
- Model Evaluation
Personal portfolio / Technical journal
AI Learner.
Agent Developer.
Learning how AI works. Building agents and tools that make ideas useful.
From intent to an evaluated result.
Exploring AI, Building What Matters.
Explore the pageAbout
Learning through building.
I'm an AI learner and developer interested in understanding how intelligent systems work and how they can be applied to real-world problems. I explore large language models, agent architectures, and automation through hands-on experimentation and independent projects.
- Background
- An AI learner and developer, exploring intelligent systems through hands-on experimentation.
- Interests
- Large language models, agent architectures, and the tools that connect ideas to practical work.
- Current direction
- Learning how to plan, build, and evaluate agents while developing independent tools.
Focus Areas
Three connected areas of exploration, from understanding models to making useful tools.
Agent Engineering
Exploring agent architectures, tool use, planning, memory, and multi-agent collaboration.
- Agent Architecture
- Tool Calling
- Task Planning
- Memory
- Multi-Agent Systems
AI Automation
Exploring AI tools, automated workflows, Python, and developer tools.
- AI Tools
- Workflow Automation
- Python
- Developer Tools
Selected Projects
A closer look at the tools I am building. Small in scope, grounded in practical problems.
A cross-platform Go CLI for listing files, computing hashes, and batch renaming. Renames preview changes by default; duplicate detection and cleanup remain planned.
- The problem
- Repeating file searches, integrity checks, and naming changes across directory trees.
- Updated
- Stars
- 0
- Forks
- 0
- Primary language
- HTML
More on GitHub
Public repositories, updated daily. Forks and archived repositories are excluded.
No additional public repositories yet.
Browse repositories on GitHub ↗GitHub data checked:
Agent Lab
A conceptual view of how an agent might move from a request to an evaluated result. This diagram describes an architecture, not a running system.
ConceptOpen a node to explore its role.
User Input
Capture the task and its constraints.
- Input
- A request and relevant context.
- Output
- An explicit goal with constraints.
- Design note
- Validate required information before planning.
Agent Planner
Break the goal into manageable actions.
- Input
- The goal and available capabilities.
- Output
- A proposed sequence of actions.
- Design note
- Planning should respect scope and stop conditions.
Memory / Context
Select information relevant to the current step.
- Input
- The plan, conversation, and prior observations.
- Output
- A focused working context.
- Design note
- Retain useful context without assuming permanent memory.
Tool Calling
Use a tool to perform a bounded action.
- Input
- A selected tool and validated arguments.
- Output
- A tool result or an explicit error.
- Design note
- Check permissions and handle failures before continuing.
Evaluation
Compare the result against the original goal.
- Input
- Tool results and success criteria.
- Output
- An assessment: finish, revise, or stop.
- Design note
- Use explicit criteria rather than assuming a successful tool call solves the task.
Final Output
Present the result and its limits.
- Input
- Evaluated findings and supporting evidence.
- Output
- A useful answer with clear limitations.
- Design note
- Distinguish verified results from assumptions and unfinished work.
Notes & Explorations
A space for technical notes, experiments, and things learned along the way.
A static React portfolio: building pages and deploying to Pages
Notes from building this bilingual portfolio: static rendering, a smaller JavaScript bundle, and two deployment configuration mistakes.
Read note
Contact
Let's connect.
Open to ideas, collaboration and conversations.