Referrals and recommendations

North-star seats are referral-heavy. Cold apply still works sometimes. It is not a strategy. I started warmth in the war room. This page is how I behave once I’m on a seat — low ego, high homework, make it easy for the other person.

Cadence

Track on the weekly scoreboard.

How referrals actually work

A strong referrer needs three things:

1. Low reputation risk — they believe I can pass a phone screen
2. Low effort — job link, resume, job ID, paste-ready note
3. A reason to care — shared employer, school, OSS, Substack, mutual friend, real technical overlap

“Hi, I saw you work at Google, can you refer me?” puts all the work on a stranger. Reply rates stay in the single digits. Messages that name a specific team, a mutual signal, and one concrete ask convert.

Highest-converting path: ask for advice → be excellent on the call → they offer the referral. Mid-tenure ICs on the target team often help more than famous Staff+ folks drowning in asks.

Target map (fill offline)

Do not commit private contact lists with emails to git.

Refresh weekly. One warm thread that progresses beats ten cold applications I forget to follow up.

Messages I’d actually send (~150 words)

A — Mutual signal → advice first (when cold-ish)

Hi [Name] — we both [shared signal: Lucid / Volvo / Kaiser / LJMU / Substack thread on X]. I’ve been following your notes on [specific topic]. I’ve been shipping governed agents and access-aware RAG — Lucid 0→1, plus a public spine a panel can click. I’m aiming at [team / problem space]. Would you have 15 minutes on what [Company] actually selects for at Staff+ on that team? Totally fine if timing’s bad.

B — Warm intro ask (they already know the work)

I’m on [seat] with ownership of governed agents / RAG / evals, and I’m building toward Principal / Staff AI Platform seats at labs and FAANG AI orgs. Would you be open to a short intro to [Name / team]? I can send a 3-line blurb plus a technical review so forwarding is easy.

C — Referral package (only after they said yes)

One message:

1. Job link + job ID
2. PDF resume tailored to that JD
3. Paste-ready internal note
4. “Should take ~2 minutes in the portal — no worries if it’s not a fit after all.”

3-line forwardable blurb

Venkata Peetla — AI platform engineer. Lucid 0→1 supply-chain AI: $10M TrueCommerce licensing eliminated, $7M annual net; Kafka/MSK; multi-rail payments. Public spine a panel can click: gateway, orchestration, access-aware RAG, eval CI. Hire: https://venkat-ai.com/hire · Technical review: https://venkat-ai.com/technical-review

Optional fourth line for Applied / lab roles

Strong on discovery → wedge → HITL for irreversible tools, and on evals that catch false confidence — not demo theaters.

What I ask on the advice call

Skip anything the careers page answers.

1. What failed candidates looked like in the last loop — isolation? evals? coding? values?
2. Whether the posted level matches how the team actually levels
3. Whether Applied / platform / research-adjacent is the right door for my profile

End by making them successful: offer a one-paragraph summary they can forward if they volunteer a referral. I never corner them with “so will you refer me?” on minute fourteen if the chemistry isn’t there.

Close the advice call:

This was useful — especially [specific]. If a referral ever makes sense, I can send a job ID and a paste-ready note so it’s two minutes. No pressure either way.

12-hour thank-you:

Thanks for the time today. I took [specific] away. Here’s the technical review if it’s handy: https://venkat-ai.com/technical-review.

Recommendation refresh (kill the iOS-first impression)

Public LinkedIn recommendations that only praise iOS/mobile quietly downlevel me on Principal AI screens. I ask people who can speak to:

1. AI architecture / multi-agent / RAG / governance
2. FDE-style embed with business stakeholders
3. Production outcomes (savings, reliability, org standards)

Ask I’d send

Would you be willing to update or add a LinkedIn recommendation focused on AI platform / architecture work we did together — or that you’ve seen in my public stack? One short paragraph on [governance / RAG / 0→1 / a standard other teams reused] helps hiring panels more than mobile-era notes. I can send three bullet reminders of projects. You write it in your voice. I don’t want ghostwritten praise.

Themes that read as real: ambiguity → ADR → shipped system; HITL / policy before irreversible actions; access-aware retrieval or eval discipline that failed a build; mentoring / bar-raising across teams.

If they ask for more names than I have:

I’d rather send two people who can speak to governance or RAG than five mobile-era quotes. Quality of the voucher matters more than a wall of recommendations.

Lab-specific honesty (same brand, different pressure)

If I’m aiming Applied / forward-deployed shaped seats:

• Talk safety and irreversible actions without reciting a blog post
• Prefer “here’s where I’d refuse to automate” over “responsible AI is important to me”
• Eval design from a vague customer ask beats architecture tourism
• Values / judgment rounds punish rehearsed certainty — practice updating mid-conversation

Same brand. Slightly different deep dive from the same capability map. No per-company pitch card.

Privacy

• Respect non-solicit / moonlighting on the seat
• No confidential customer names or unreleased Lucid IP in public asks
• Recruiter threads and offer details stay offline
• Don’t ask current teammates to refer me to a competitor in a way that burns them

“How are you running your search?”

Referral-first. I’m doing the homework on specific teams, asking for short advice conversations where there’s a mutual signal, and only then asking for a referral with the job ID and a paste-ready note. I’m not spraying cold apps — at this level the hit rate isn’t worth the noise, and I’d rather protect people who vouch for me.

Related

• Stabilize
• Close or upgrade
• Pitch