Marcin Borecki · Software architect · Microsoft Regional Director
Engineering that holds up in production.
Deep, tested articles on .NET, Azure, databases and AI agents. Each one shows the mechanism, gives you an example you can run and names the trade-off you will have to make.
Topics
Six fields. One standard of evidence.
Real systems cross boundaries, so most articles connect more than one field. Pick a starting point.
AI Engineering
Agents, RAG and LLM features that survive production: contracts, evaluation, authority, cost and recovery.
54 articles.NET & C#
Runtime behaviour, async, memory and APIs, explained from the source and checked with code you can run.
53 articlesPerformance & Data
Databases, telemetry, ingestion and the measurements behind a system that is actually faster.
37 articlesAzure & Architecture
Identity, deployment, events, recovery and the boundaries that decide how a cloud system behaves.
19 articlesEngineering Practice
Testing, reviews, diagnostics and the habits that keep software changeable as teams and tools change.
19 articlesAI, Work & Business
What AI changes for teams, products, careers and investment decisions, argued with evidence rather than hype.
16 articlesRecently published
New in the notebook
Async Work Is Not on the Thread Stack: Tracing Suspended .NET Code
A checkout hangs at an await while every thread looks idle. Find it on the heap with dumpasync, fix the race, and measure what dumps and traces cost a live .NET 11 service.
Why .NET 11 Removes This Box, but Keeps That Allocation
One binary on .NET 10 and .NET 11: nullable boxes that stay local stop allocating, escaping boxes and closures do not. Read the JIT code and the measurements behind each case.
An AI Investment Thesis Should Change When the Evidence Does
Read an AI startup's thesis as a ledger of claims: why selected demos are not samples, which denominator a unit cost hides, and how to tell honest narrowing from moved goalposts.
From AI Draft to Accepted Work Order: One Service, Real Review
A complete-looking AI draft can pass every validator and still overstate its evidence. Route forms and IChatClient through one service and bind approval to revision and evidence.
When Code Gets Cheap, the Scarce Part Is the Promise Around It
AI cuts the cost of a first version, not of checking, upkeep and exit. An effort model, the measured speed-ups and three products built on one codebase show where value moves.
A Minimal .NET Agent Platform: Four Jobs, Two Workers, One Restart
Stop a small ASP.NET job service while it still owns work, then restart it. Four jobs show what can resume, what must wait, and what only the receiver's receipt can settle.
Series · 10 parts
Decision APIs: LLM features you can defend
From one small judgement to probabilities, thresholds, human review and idempotent resumption, implemented in .NET and Azure.
Open the series- Decision APIs in .NET: One Judgment Inside a Working Support Desk
- What a Decision Probability Actually Computes: Logits to Labels
- Designing a Decision Contract That Preserves Meaning in C#
- Decision APIs: Derive the Threshold From the Cost of Each Error
- Native Decision API Clients in C#: Wire Shapes and Stalled Bodies
Beyond the articles
Leading change, backing founders
The same rigour as the writing, applied where decisions are made: in interim executive roles, on boards and as an investor.
Interim CTO, CEO and board roles
Restructuring products, departments and IT, and building products, with a real mandate and in cycles of six months or more.
Work with meBusiness angel and VC
Tickets from €25k to €150k for technical founders in healthcare, telemetry and big data, and AI.
InvestingWorkshops for engineering teams
Hands-on training on async .NET, performance, databases, Azure and production AI, built around your own code.
Training