HeyJaredTake It From TechLead
Actionable advice for engineers to grow intentionally and become an effective leader in 5 mins. Written by Staff Software Engineer @OfferUp | Ex Msft
- Indexed issues, last 90 days
- 3
- Latest publication
- Jul 30, 2026
- Audience
- Checking…
- Earliest in this view
- Jul 13, 2026
Latest issues
You Are Not Paying Claude to Think. You Are Paying It to Re-Read (opens the original)
Read excerpt
Day 11. My Claude plan hit its monthly limit.Not slowed. Done. The meter said I had spent a month of AI in a week and a half.My first instinct was to buy a bigger plan. But something bugged me. I don’t type that much. My prompts are short. My files are small. Where did a month of tokens actually go?So I followed them. All 172 million of them.The answer changed how I use these tools every day. About 92 of every 100 tokens were the model re-reading things it had already read. Only 1 in 100 was the
We Argued About Dinner Every Single Night. So I Built an AI Agent to End It (opens the original)
Read excerpt
At 5 PM, our Telegram group buzzed.Tonight’s dinner. The dish, the ingredients, one prep note. Nobody typed it. Nobody had to remember it.Saturday morning, it buzzed again. The week’s menu, plus the full grocery list, sorted by aisle.Last week I texted the bot: “change Wednesday dinner to rajma chawal.” It replied “Menu updated.” That was the whole deploy.And behind all of it? No server. No database. No app.An AI agent I built in one evening. $0 a month.This is the full build log. Every decision
How I Built an AI Job Agent That Never Sleeps (opens the original)
Read excerpt
At 9:00 AM, her phone buzzed.An AI had read 157 job posts overnight.It picked the best five, scored them, ranked them, and sent them to her on Telegram.She replied:“Apply to #3.”Ten minutes later, a tailored two-page resume landed in her chat as a clean PDF.A friend of mine is job hunting.I built this for her in one day.I’ve been an engineer for 12 years. This is the fastest I’ve ever built anything. AI wrote most of the code.And that speed hides a trap. This post is about the trap, and the way
Publishing over time
Last 90 days. Choose a month to open its work.
Recurring subjects
Named in the text we hold. One piece can cover several.
Audience
No verified audience measurement yet.
About this data
Counts cover the work we have indexed. Tone needs enough text and a confident classification. Excerpts and episode notes are not full articles or transcripts.
Identity or attribution wrong? Suggest a correction.