LinkedIn Growth Tracker

Matthew Lopez 1st · ex-xAI · Multimodal AI specialist

"One year ago today, I started working at xAI… 10 things I learned about resilience, growth, layoffs…"

Matthew, the "record your wins" one hits hard. I started doing that after my role shifted at xAI and it changed how I interview — instead of describing duties I now point to specific shipped work. Point 7 is underrated too: most engineers I know undersell their work because the abstraction is so internalized they don't realize how much judgment went into it. Glad you kept going.

Malcolm Matalka 3rd+ · Stategraph

"REST's dead, baby. Just do SQL queries direct from the client… we built a sandboxed SQL interface so frontend devs write their own queries."

Open post →

The dashboard-as-server-change problem is real — I hit the same wall building an admin portal recently and ended up with a half-dozen bespoke aggregate endpoints. Curious how you handle authorization in the sandbox though: is it row-level policies in Postgres, a schema rewrite layer in your parser, or both? Letting frontends write SQL is great until someone selects a column they shouldn't see.

Lucas Cabral 3rd+ · AI & Software Lead, SitecomAI

"Everyone wants to become an 'AI Engineer' now. But most people still don't understand basic web architecture… prompting is not engineering."

Open post →

Agree on the foundations point. The "AI app" category has flipped what used to be a backend problem (orchestration, retries, idempotency, observability) into the hard part of the product — except now the failure modes are non-deterministic. Prompting is the cheap layer; the engineering is everything around it that keeps a flaky model behaving like a reliable service.

Alex Cinovoj 3rd+ · 4M+ impressions/90d

"Job post required 4+ years of FastAPI. He couldn't apply — he only has 1.5+ years. Since he CREATED the thing."

Open post →

Worst one I've seen: "5+ years of Next.js App Router" — App Router shipped in 2023. The pattern says more about the recruiter's understanding than the candidate's. The fix isn't AI screening, it's letting hiring managers write their own requirements based on what the job actually needs.

Anton Zaides 3rd+ · EM @ HoneyBook, manager.dev

"AI is pushing engineering managers back to building. I made this exact mistake 7 years ago… balcony awareness."

Open post →

The "stepping off the ladder" framing is the cleanest description of the IC→EM transition I've read. As an IC watching this shift, I'm curious about the inverse risk too: do you think there's a version of "balcony awareness" engineers should be practicing now that AI is doing more of the rote coding? The pull to over-rely on the agent feels structurally similar to an EM diving back into the IDE.

Post Pipeline

Week 1 (publish Wed May 20): Ready ADVEC TypeScript PostgreSQL

I built an admin portal for my church. Somewhere along the way I ended up implementing geofencing to automatically assign pastors to home visits. The feature: when a new household visit gets scheduled, the system checks the address against geographic zones and assigns the nearest pastor and volunteer team. No manual lookup, no spreadsheet. The interesting part was that it's not just about distance. Each zone is defined by a radius around a central coordinate. I wrote a query that calculates the straight-line distance between the visit address and each zone center, then picks the closest match. If no zone covers the address, it falls back to a default team. The whole thing runs on TypeScript, React 19, Express 5, and PostgreSQL, deployed on Hetzner Cloud via Docker Swarm. It's live at portal.advecboston.com. What I took away from it: the most interesting engineering problems don't always come from big companies. Sometimes they come from real people with real workflows that no off-the-shelf tool handles well. Fellow engineers, what's the most unexpected feature you've ever had to build?

Week 2: Idea SouthCoast Wind Automation

Hook: "I spent 8 months at an offshore wind company replacing 80–90% of a manual document workflow with Power Automate and Python. Here's what surprised me." Beats: • What the workflow actually was (regulatory compliance docs — lifecycle, audit trail, sign-off) • Why manual was the default (compliance fear: people trust signatures, not scripts) • The hardest part wasn't building it, it was getting people to trust the output • Closing thought: "Automation in regulated industries is 20% code, 80% earning the right to remove a human from the loop." CTA: Anyone else automated work in compliance-heavy spaces? What did it take to get sign-off?

Week 3: Idea ADVEC Docker Swarm Self-hosted

Hook: "I run my church's app on a $5/mo Hetzner box with Docker Swarm. Here's why I didn't reach for AWS." Beats: • The case for self-hosting when you're the only operator • What Docker Swarm gets right that Kubernetes overengineers • When the savings stop being worth it (the day you have to page yourself at 3am) • What I'd do differently if traffic doubled CTA: For solo-dev side projects, where do you land — managed (Vercel/Fly), self-hosted, or something in between?

Week 4: Idea DroneAR Career

Hook: "Two years ago I was a document control manager. Today I'm building drone+AR systems and shipping production code. The pivot wasn't a bootcamp." Beats: • What I was actually doing as document control (and why it gave me ops/process intuition) • How the Harvard CS degree helped — and what it didn't teach • What unlocked the move: shipping things that I cared about, not building portfolio projects • What I'd tell anyone making a similar pivot CTA: To anyone who pivoted into engineering — what was your "this is actually real" moment?

Connection Targets

Matthew Lopez 1st

ex-xAI · Multimodal AI Specialist

[Already connected — DM instead] Hey Matthew, your 1-year reflection post hit. I'm trying to build out my LinkedIn presence after the xAI chapter and would really appreciate a recommendation if you have a few minutes. Happy to write one back. No rush.

Open profile →

Recommendation target

Andy Acevedo 3rd

xAI · Harvard I-Labs member · Boston-adjacent

Hi Andy — fellow xAI/Harvard person here. Saw you're at the I-Labs too. Would love to connect.

Open profile →

Daniel Kretschmer 2nd

Human Data @ xAI / SpaceXAI

Hi Daniel — also did Human Data at xAI before moving more into engineering. Would be great to be connected.

Open profile →

Sydney Wendling 2nd

Human Data Team Lead @ xAI (Post-Training)

Hi Sydney — was on the Human Data side at xAI before moving into engineering. Would love to connect; you've probably worked with people I know.

Open profile →

Sota Ikeda 3rd

AI Research Scientist · xAI alum

Hi Sota — fellow xAI person. Building backend/full-stack things post-xAI; happy to be connected.

Open profile →

Saketh Mynampati 3rd+

SWE @ Axiomatic_AI · Harvard SEAS MS · Cambridge MA

Hi Saketh — Harvard CS person also in Cambridge, working on backend/full-stack. Would be great to connect with another local in the field.

Open profile →

Next Steps Beyond This Week

Re-check SSI in 2 weeks — visit linkedin.com/sales/ssi. Expect the biggest jump on the "Engage with insights" pillar (currently 0.45) since 5 thoughtful comments per week is a 10x increase from your baseline.
Featured section — pin portal.advecboston.com, the AWS Cloud Practitioner badge, and the geofencing post once it goes live.
Profile photo & banner — if either is more than 2 years old, refreshing one is a measurable bump on the "professional brand" pillar.
Newsletter or article — at the 500-connection mark, LinkedIn unlocks Creator Mode features that compound. The geofencing post + 2 follow-ups is enough material to launch a short engineering newsletter.

LinkedIn Growth Tracker

SSI Scorecard

Pillar breakdown

This Week's Habits

Engage with Insights — Drafted Comments

Post Pipeline

Connection Targets

Next Steps Beyond This Week