Book → capstone → solution → repository

Solutions

Start with a real AI problem. Follow it through the mechanism, evidence, capstone implementation and inspectable code to a solution you can actually use.

Programmer.ie solutions are the practical destination of the research, books and experiments.

A book should not end at explanation. Each finished book is expected to lead to a solution: a durable page organised around a real problem, a concrete outcome and an implementation that can be inspected. The book earns the mechanism, the capstone integrates it, and the solution becomes the applied home for the work.

The publishing loop

    flowchart LR
    R[Research] --> B[Book]
    B --> C[Capstone]
    C --> S[Solution]
    S --> G[Repository]
    G --> U[Real use]
    U --> E[New evidence]
    E --> R
  

A solution can also begin somewhere else. A strong blog post, experiment or engineering investigation can grow into a solution when it becomes comprehensive enough to answer the problem rather than merely discuss it. The solution page then becomes the canonical home, while supporting posts provide narrower entry points into it.

How to use this page

  1. Start with the problem Browse the solutions below by domain or featured project.

  2. Understand the mechanism Each solution connects back to the book, experiment, or underlying technical argument.

  3. Inspect the implementation A real solution leads to code, experiments, tools, or a live system.

What a solution is

A solution on Programmer.ie is not just an article and not just a repository.

It is the place where:

  • a real problem is named clearly,
  • the mechanism is explained properly,
  • the capstone implementation is gathered,
  • the code or live system is linked,
  • and the result can be inspected, tested or used.

Every mature solution should make five things easy to find:

  1. The problem — what is failing, missing or unnecessarily difficult?
  2. The mechanism — what technical idea actually determines the outcome?
  3. The evidence — what measurements, experiments or source material justify the design?
  4. The implementation — what capstone, repository or repository path embodies the answer?
  5. The operating boundary — what the solution does, does not do and how you can tell whether it is working.

Some solutions below are live systems. Some are being built directly from current books. Others will appear as the library grows. The goal is not to manufacture a product for every chapter. It is to make sure important technical work eventually terminates in something concrete that can be built, tested, inspected or used.

01Book

Understand the current technique, mechanism, evidence and failure modes.

02Capstone

Turn the argument into a complete implementation that can be tested.

03Solution

Package the result around a real problem and a concrete outcome.

04Repository

Keep the implementation, experiments and evidence inspectable.

22Total solutions
5Live
1Building
16Reference
0Planned
Solution library

Browse by domain

Browse the solution library by domain. Each solution page connects a real problem to its mechanism, implementation and supporting evidence.

Agents

4 solutions
Reference Component

Ollama Python Tool Calling

Expose local Python functions to an Ollama chat model and execute returned tool calls through an explicit registry.

Problem Local LLM applications need a controlled way to call real Python functions instead of relying on text-only responses.
Outcome A small function registry and execution loop for Ollama tool calls.
Reference Agent

LATS Reasoning Tree Search

Use Monte Carlo Tree Search, dimensional scoring, and symbolic refinement to explore multiple reasoning paths.

Problem Greedy single-path agents often commit too early and provide little evidence about why one reasoning path was better than another.
Outcome A Language Agent Tree Search loop with structured state, UCT selection, dimensional scoring, and trace-based refinement.
Reference Agent

Case-Based Plan Reuse for AI Agents

Retrieve, rank, adapt, validate, and retain past agent plans so new runs can reuse proven reasoning without fine-tuning.

Problem Agents usually start each task from scratch even when previous runs contain useful plans, failures, and successful reasoning traces.
Outcome A reusable memory layer that selects high-quality prior cases, adapts them into a new plan, validates the result, and records parent-child reuse links.
Reference Agent

Search-Solve-Prove Self-Play Loop

Generate questions, solve them with search, verify answers against evidence, and turn each episode into metrics for self-improvement.

Problem Agent self-improvement needs generated practice tasks, but generated answers are useless unless evidence and verification are part of the loop.
Outcome A repeatable SSP episode format with evidence, verification, normalized metrics, visual frames, and replayable traces.

Ai Memory Infrastructure

1 solution
Live Plugin

OpenCode Remembering

A standalone memory runtime for AI agents that separates retrieval, historical recall, temporal validity, current work, trust, decisive selection, traceability, unfinished work, and explicit memory writes from permission to influence present behaviour.

Problem Long-running AI agents accumulate files, conversations, decisions, failures, instructions and changing project state, but ordinary retrieval cannot tell historical evidence from current guidance, relevant material from authoritative material, or remembered information from permission to act on it.
Outcome A standalone remembering runtime where canonical project history is retrieved strongly, routed by purpose, interpreted temporally, conditioned on the work being done, gated by standing, reduced to decisive provenance-bearing evidence, recorded in an immutable trace, with unfinished work tracked and explicit memory actions accepted without granting the writer authority over behaviour.
Book: Memory From First Principles Capstone: The Memory From First Principles application: a real coding-agent plugin implementing the book’s earned path from canonical project history through retrieval, routing, temporal state, safe framing, trust, decisive selection, durable traces, open loops, and explicit memory actions.

Applied Ai Infrastructure

1 solution
Live Runtime

CodeAI

A durable runtime for putting model intelligence inside an inspectable process with explicit state, authority, effects, verification, acceptance, replay, and deterministic next-operation policy.

Problem A model call can produce useful intelligence without producing a reliable process: chat history hides state, successful requests are confused with completed work, model claims become evidence, retries repeat side effects, and the system has no durable answer to what should happen next.
Outcome A runtime where stochastic cognition is bounded by deterministic state projection, explicit authority, preserved observations, verification, acceptance, replay, and a recorded policy for choosing the next operation.
Book: Applied AI Capstone: The Applied AI capstone: one task carried from directive and context through model cognition, controlled effect, bound verification, acceptance, replay, and an audit of the joints between them.

Automation

1 solution
Reference Pipeline

Research Paper Video Generation Pipeline

Turn research narration into timestamped transcript chunks, image prompts, generated images, and an FFmpeg video.

Problem Making research videos by hand requires transcription, chunking, visual planning, image generation, and video assembly.
Outcome A scriptable pipeline that stores transcript segments in SQLite, generates visual prompts, creates images, and assembles a video with FFmpeg.

Browser Ai

1 solution
Building Tool

Digital Lens

A user-owned browser AI layer that can filter, amplify, search, organize and act on information while keeping context, provenance and authority inspectable.

Problem The browser contains the user’s real information environment, but today context is fragmented across pages, tabs, messages and services while AI assistance is usually trapped inside isolated provider interfaces.
Outcome A user-controlled browser layer that can find, filter, amplify, retain and act on information using explicit policies, typed tools and inspectable evidence.
Book: Browser AI From First Principles Capstone: A user-owned browser AI coordinator and Digital Lens built on the Browser AI Observatory, typed tools and explicit authority boundaries.

Build Tools

1 solution
Reference Utility

Density-Aware Asset Rasterization

Generate Android density-specific PNG assets from source SVG files with one repeatable command.

Problem Mobile projects need repeatable raster assets at several pixel densities, but manual export creates inconsistent sizes and stale files.
Outcome A small command-line script that maps SVG names to Android resource folders and renders the required PNG sizes.

Data

3 solutions
Reference Pipeline

Document Intelligence Pipeline

Retrieve, parse, classify, score, and select research documents for goal-aware AI systems.

Problem Research agents need to know which documents are relevant, reliable, and useful for a current goal.
Outcome A pipeline of agents that loads documents, profiles sections, classifies domains, scores papers, and selects goal-matched knowledge.
Reference Utility

Schema-Driven File Validation

Validate CSV or spreadsheet exports against a declarative schema before importing them into another system.

Problem Data exports often fail only after import because required columns, dates, numbers, allowed values, or field lengths were not checked first.
Outcome A reusable validator that reports row, column, and reason for every schema violation before the file reaches the destination system.
Reference Component

Immutable Key-Value Lookup Store

Build a compact read-only lookup table for fast local key-value reads without running a database service.

Problem Some applications need fast local reads from reference data that changes rarely, where a full database engine is unnecessary.
Outcome A simple immutable store that writes key offsets once, then performs direct reads by key from a local binary file.

Embedding Infrastructure

1 solution
Live Runtime

RELATE

A relation-aware embedding runtime that measures identity, compatibility, preservation, and policy-scoped usability instead of trusting raw vector similarity.

Problem Embedding systems often treat proximity as permission: if two vectors are close, the application behaves as though meaning, equivalence, compatibility, or preservation has been established.
Outcome A measured runtime that records space identity, evaluates geometry and transformations, preserves evidence, and grants usability only for scopes that passed declared tests.
Book: Embeddings From First Principles Capstone: RELATE 1.0: the embedding runtime, corpus, benchmarks, preservation profiles, and replayable evidence behind the book.

Evaluation

2 solutions
Reference Pipeline

Model Gap Calibration and Routing

Compare two models through shared metrics, measure disagreement, calibrate the smaller model, and route uncertain cases to the stronger model.

Problem A cheaper model may be good enough for many cases, but raw scores do not show where it disagrees with a stronger model or when to escalate.
Outcome A repeatable pipeline that aligns model outputs into shared metrics, computes delta fields, calibrates scores, and emits routing policy.
Reference Pipeline

Visual Epistemic Field Generation

Convert scored reasoning traces into Visual Policy Maps and differential epistemic fields for debugging AI reasoning.

Problem Reasoning scores and logs are hard to inspect at scale, especially when the useful signal is a pattern across many traces.
Outcome A reproducible process that turns metric matrices into visual fields with provenance and metric-level explanations.

Evidence and Reliability

1 solution
Live System

ZeroModel

Evidence-first AI systems where claims, decisions, policy boundaries, failure states and replay remain explicit.

Problem AI systems can cross from model output into decisions and actions without preserving the evidence, claim boundaries and policy state needed to audit or replay them.
Outcome A bounded decision system whose important claims, evidence, decisions, failures and replay state remain explicit across runtime boundaries.

Human Ai Cognition

1 solution
Live System

AIbussin / Freestyle Cognition

Practical techniques and interactive tools for using AI as a thinking, exploration and creative partner rather than a one-shot answer machine.

Problem Most AI interfaces encourage one-shot prompting even when the real task is exploration, reframing, comparison, synthesis or creative development.
Outcome A reusable set of human + AI thinking techniques that can be applied as structured workflows instead of improvised prompts.
Book: Freestyle Cognition Capstone: Turn cognition techniques into reusable interactive workflows and tools.

Language Model Programming

1 solution
Reference System

Self-Improving Engineering Program

A DSPy-backed repository-repair loop that turns language-model program changes into candidates evaluated by independent evidence before promotion.

Problem Language-model programs can optimize toward convenient feedback while confusing a better score, a valid patch and a trustworthy production change.
Outcome A controlled repair loop where the LM can diagnose and propose, but ordinary software applies patches, checks scope, withholds independent evidence and records a separate promotion decision.
Book: DSPy From First Principles Capstone: Chapter 20 repository-repair capstone: repair-run-20260901-013940.

Rag

2 solutions
Reference Component

FAISS and SQLite Vector Search

Persist text chunks in SQLite while using FAISS for fast local similarity search over Ollama embeddings.

Problem FAISS is fast at vector search, but applications still need durable document metadata, text chunks, and lookup tables.
Outcome A reusable wrapper that stores embedding provenance in SQLite and uses FAISS to retrieve the matching document text.
Reference Component

Hybrid RAG Search with sqlite-vec

Combine SQLite full-text search with sqlite-vec semantic retrieval and Reciprocal Rank Fusion.

Problem Keyword search and vector search each miss useful results in different ways.
Outcome A local hybrid retriever that combines lexical and semantic matches without running a separate vector database service.

Research Automation

1 solution
Reference Pipeline

PaperSearch arXiv Retrieval Pipeline

Search arXiv, download paper PDFs, extract text, chunk pages, and store the results in SQLite for RAG and research automation.

Problem Research agents need repeatable access to papers as structured local data, not one-off downloaded PDFs.
Outcome A SQLite-backed paper retrieval and processing pipeline that can feed semantic search, RAG, and downstream research agents.