Join The Modern Engineer Newsletter

Build a deterministic context budgeting pipeline that prioritizes required evidence, enforces token limits, and composes inspectable prompts.

Use structured JSON inputs and outputs to make LLM systems predictable with typed parsing, validation, and deterministic routing.

Learn the core AI pattern for semantic retrieval with embeddings as representation, deterministic vector ranking, thresholded control, and optional pgvector scaling.

Build a local-first incident triage agent with Microsoft Agent Framework using structured workflows, deterministic policy enforcement, and explicit runtime guardrails.

Spec-driven AI-assisted coding treats specifications as execution contracts, constraining agent behavior for predictable, production-grade outcomes.

Build a local-first .NET intent router that keeps routing, tools, and safety deterministic while using embeddings only for disambiguation under controlled thresholds.

Build an evaluation harness with telemetry and feedback loops to gate local AI releases with measurable reliability. Track quality over time, catch regressions early, and ship with clear deployment thresholds.

Build a minimal but production-aligned guardrailed AI assistant in C# that runs fully locally using Ollama, focusing on clear boundaries, strict contracts, deterministic execution, and safe failure modes.

Build a local semantic search engine for engineering runbooks using Ollama and Microsoft.Extensions.AI for deterministic retrieval.

Build a reactive local AI agent using Semantic Kernel and Ollama for deterministic tool execution and natural language reasoning.