Wolven Tech: Rust consulting. AllSource Prime: durable AI memory → https://xitter.cf/t.co/CfdDxoKhs5 Longhand: AI workflows & training → https://xitter.cf/t.co/Rbml5TOtFF
London, United Kingdom
Joined January 2014
- Tweets4.5K
- Following2.8K
- Followers1.2K
- Likes9.3K
Pinned Tweet
1,000 followers. Thank you.
Marking it by opening something up: I am taking on fractional full stack AI developer work through Wolven Tech Ltd.
Five spots. One is already taken.
How it works below.
A buffered log can be written after its request context changes. OBI's trace-log correlation report warns that asynchronous writes under backpressure can miss context or carry stale context. 1/5
🤖 Made with AI
ALT Red and silver shapes pass through a dark buffer onto a conveyor. A lamp illuminates their geometric identifiers at the outlet, watched by a sculptural silver wolf. Caption: Check the write boundary.
Delay output in a representative service and interleave requests. Compare emitted records with the requests that created them. Inspect after the buffer; a log-creation check can miss the later write context. 4/5
This is a proposed qualification test, not a measured failure rate. Retain logging configuration, covered paths and known gaps so investigators can recognise unavailable correlation.
opentelemetry.io/blog/2026/o…
5/5
LightOn's OCR report describes normalising equivalent representations before scoring. Footnote formatting can change an edit-distance penalty while leaving the extracted information intact. 1/5
🤖 Made with AI
ALT Round and diamond-faced silver stamps sit above matching pale broken-circle impressions. A separate red impression remains different; a wolf seal rests nearby. Caption: Version the normaliser.
Build paired fixtures: equivalent content expressed differently, plus changes that alter information the application needs. Require the latter to remain distinguishable. Keep the fixtures with each normaliser revision. 4/5
Before using an OCR score to choose a model, identify the normaliser revision and inspect raw output for an example whose score changed. No independent ranking is established by this review.
huggingface.co/blog/lightona…
5/5
7/11 Jian describes blanking on JS syntax and recursion, then predicts engineers will stop reading code. I disagree with that prediction.
Allow syntax lookup. Still ask candidates to trace the recursion and explain why it terminates.
xitter.cf/jianxliao/status/20294…
Yesterday I had my first coding interview, at one of the big AI labs, after 4 years of being a founder.
It was a disaster.
The task? agents algorithm. something I work with literally every single day.
I forgot basic js syntax. blanked on how to delete an array element. panicked on recursion.
The solution was crystal clear in my head. I could see exactly how to write it. but my hands just... couldn't.
The knowledge is there. the muscle memory is gone.
3 years of vibe coding did this to me. I haven't written code manually since. I just read it, design systems, think in architectures.
Somewhere along the startup journey, I stopped being a coder. I became someone who just ships.
Am I alone in this?
Sitting there, embarrassed, I think that's actually the right direction.
We used to write code. now we read it. soon with agentic engineering, we won't even need to read it. we'll just architect.
10/11 Sterling promotes AI-cheating detection. I haven’t verified his accuracy claims.
Set explicit rules for permitted AI use and identity. Candidates should know how to challenge a disputed flag. A flag needs review before a hiring decision.
xitter.cf/SterlingSmith/status/2…
6/11 Lior proposes building with Claude Code while an interviewer watches.
Include a review of what the tool got wrong. A generated test can share the code’s mistaken assumption. Ask the candidate to challenge it.
xitter.cf/liorpesoa/status/21140…
4/11 Predict the rows before running the JOIN: Nikita suggests this SQL exercise.
Add duplicate keys and NULLs. Allow docs or AI, then ask the candidate to explain an inflated count and demonstrate a correction.
xitter.cf/nikita_vidiakin/status…
1. Basic SQL
Let's start with the obvious one.
SELECT, WHERE, ORDER BY, LIMIT, DISTINCT
You don't need to think about syntax here. These things should already feel natural when you start solving interview tasks.
2. JOINs
You will definitely meet them.
At least be comfortable with INNER JOIN and LEFT JOIN.
But don't just remember the syntax. You should understand what actually happens to the rows after your join.
Take two small tables and try to guess the result before running the query. It helped me a lot.
3. Aggregation
GROUP BY, COUNT, SUM, AVG, MIN/MAX, HAVING
Also make sure you understand the difference between WHERE and HAVING.
And yes, get used to things like:
COUNT(DISTINCT user_id)
You will see them all the time.
4. Subqueries & CTEs
You don't need to make every query a huge construction with five nested subqueries.
Just understand how they work and when splitting one big problem into a few smaller ones makes your life easier.
Same thing with WITH / CTEs.
5/11 Danny Thompson argues that candidates should explain decisions behind their projects.
I’d change one requirement and ask them to update the code and tests. Follow the explanation with something the interviewer can inspect.
xitter.cf/DThompsonDev/status/21…
One of the biggest mistakes developers are making in the 2026 job market is confusing a finished project with proof of engineering ability.
Y'all need to understand something:
Nobody is hiring you because your portfolio has a cool application. They ARE hiring you because of the engineering conversations that application allows you to have.
Think about two candidates walking into an interview with the exact same AI-powered project.
- Candidate #1 explains what the application does, which AI tools they used, and shows off the features.
- Candidate #2 explains WHY they designed it the way they did. Where they used deterministic logic instead of AI. How they handle unreliable model outputs. What happens when the system fails. What tradeoffs they made and WHY.
Same exact project. Two completely different interviews.
One demonstrated that they could produce software.
The other demonstrated that they understood how to engineer it.
And this is why I keep pushing developers to understand the decisions they're generating with AI.
I don't really care if AI wrote 100% of your code. But if you can't explain the architectural decisions, defend your tradeoffs, or identify where things could go wrong, you've missed one of the biggest opportunities that project was supposed to give you.
AI lowered the barrier to building software. It did NOT lower the bar for understanding it and making the right decisions.
We got into this on the latest episode of The Programming Podcast because I think developers need to fundamentally rethink what they're trying to accomplish with portfolio projects.
#code #coding #programming #softwareengineering
2/11 Antonis describes reviewing supplied code and preparing it for production.
I’d ask which defect blocks release and what evidence would clear it. Allow docs and AI; examine how the candidate prioritises risk.
xitter.cf/Aevmorfop/status/21128…
Nah I think it was a good interview actually.
I was given a piece of code by “an intern”, review it and productionise it (with sensible expectations given the timeframe).
They did test the knowledge rather than the syntax so it was a good experience overall.
They apparently used to have take home tasks but stopped them cause Claude 🤣
3/11 A flaky test would make a concrete work sample for Mejbaur’s argument about test architecture and reliability.
Ask candidates to distinguish shared test-data leakage from a product race. Adding retries until CI turns green leaves that unresolved.
xitter.cf/fagun018/status/211143…
1/11 Nandkishor says an interviewer wanted a complete CI pipeline from memory, with zero syntax errors.
The syntax-check interview cannot die soon enough. Ten X discussions prompted this follow-up.
xitter.cf/devops_nk/status/20110…
Had a really unexpected interview experience today.
After intro & responsibilities, they asked me to share screen and write a complete GitHub Actions CI/CD workflow end-to-end — without referring any documentation and with zero syntax errors.
I explained the entire production flow clearly:
- branch strategy
- checkout
- build
- code scan
- image scan
- push to ECR
- update manifest for Argo CD
Still, that wasn’t “enough”.
Honestly asking — who writes CI/CD pipelines, Kubernetes manifests, or Jenkinsfiles from scratch without docs ?
Even in Terraform for AWS has 200+ services. Are we expected to mug up every syntax ?
In real projects, good engineers design systems, understand flows, and refer docs not memorize YAML.
Seniors & hiring managers — is this normal, or are we confusing memorization with real DevOps skills ?