AI engineering
Smaller models. More interesting trade-offs.
A practical look at choosing a model around latency, cost, and the job it needs to do.
Follow this ideaA little help staying present
Less noise.
More you.
Relevant tech news. Posts in your voice.
A clearer path to your next opportunity.
The reading list
The world moves fast.
Follow what matters.
AI breakthroughs. Framework releases. The ideas that change how you build.
AutoLinkedIn brings public sources into one reading list, shaped around your experience and interests.
Illustrative reading listPicked for relevance
What makes an agent reliable?
Evaluations before another prompt rewrite
Example topics, not live headlines. Your private reading list uses linked public sources.
A starting point.
Not a stand-in.
A useful source becomes a first draft, informed by your CV and writing style. You give it the final word.
The most capable model isn’t always the most useful one.
For a focused task, I’d start with the smallest model that clears the evaluation bar. Then measure latency, cost, and the cases where it fails.
The interesting engineering question: what does this workflow actually need?
In the app, this draft waits
for your approval in Telegram.
Interactive example. Nothing is published.
ResearchDraftYour approval
Keep your next
move in view.
The same experience that shapes your posts can help you find relevant roles. Keep the search, the preparation, and the follow-up together.
Find the fit.
Discover roles from supported job boards and see why they match your background.
Job discoveryMake it personal.
Prepare application drafts from your experience, with room to edit every word.
Application preparationKeep the thread.
Track your pipeline, save notes, and keep follow-up dates on the calendar.
Your pipelineApplication drafts are yours to review and submit. The bot does not send them to employers.
A few good
questions.
Automation with a clear handoff.
Does it publish without asking?
No. Posts need approval of the current draft in Telegram. Editing a draft requires a fresh approval. Simulation mode lets you check the flow before publishing.
Where does the news come from?
Publisher feeds, Google News discovery, and web research can bring in relevant technology stories. Drafts use checked publisher sources, so you can review the context yourself.
Can it help with likes?
It can score posts from a configured watchlist and queue relevant likes for approval. Automatic likes require an explicit setting and the necessary LinkedIn API permissions. It does not browse your entire personal feed.
Who can open the workspace?
This is a private workspace for its configured owner. Sign in through Telegram and approve the matching code. There is no public account registration.