Hello, recruiters.

• A product where AI, security, and infrastructure all meet — and the technical bar is real.
• A team that ships and reviews each other's code, and doesn't apologize for caring about how things work.
• A founder or leader who has done this before — or is humbly figuring it out, with the receipts to back it up.
• 0→1 or 1→10 problems where I can actually shape architecture, hiring, and direction.
• Companies that respect the boundary between an "AI demo" and an "AI product."
• Roles that combine hands-on architecture with team leadership — not just one or the other.
• A CTO or founding-CTO seat at an early-stage company with real funding, a named lead investor, and a founder who has done the homework — where the first 5–15 hires and the technical direction are mine to set.

No hard feelings on these — just patterns I've learned to recognize:

• Pure people-management roles where I never see code, architecture, or a real customer again.
• Roles where "AI" is being bolted on to please the board, not to solve a real problem.
• "Stealth" companies with no team, no funding, and no concrete product to point at — including the "looking for a technical co-founder" outreach with no cash budget on the table.
• Generic full-stack roles where the AI / security / infrastructure parts are decorative.
• Web3, crypto, or "agentic-everything" pivots without an actual problem to solve.
• Compensation conversations that begin with "we'll get to numbers later."

I read offers carefully. Equity, vesting, cliff, refresh, RSU vs stock options, strike price, dilution — I know the difference and I'll ask about all of them. Bring the spreadsheet; I won't be the one who pretends it doesn't matter.

Counter-offers don't insult me. Lowball numbers do. If your range is tight, please share it up front. It saves both of us a discovery call we don't need.

I'm based in India, but I've worked alongside US, UK, and SG teams. My expectations track the role's geography and seniority — not a single global number. Happy to discuss specifics with the right context in hand.

Email is best: vishal.prince30@gmail.com. In your message, please include:

1. Subject line — company, role title, one-line about the team. ("Acme · Founding AI Engineer · ex-Google team of 4")
2. The "what" — one paragraph on what the company is solving. Pretend I've never heard of it, even if I have.
3. The range — base + equity, or at least the bracket. Yes, really.
4. The hook — one specific thing about my work, writing, or projects that made you reach out. (This is how I tell apart hand-written outreach from a sequence.)

• "Quick 30-min chat to align." — please send details first; then we'll align.
• "We can talk numbers later." — please share the bracket up front.
• "Tell me about yourself." — the rest of the site covers it in detail.
• "This is a once-in-a-lifetime opportunity." — they rarely are. The good ones speak for themselves.
• "Are you actively looking?" — the site says open to selective work. That's the answer.
• "What's your notice period?" — we'll get there if we get there. It's not the first question.

A small appendix — because the AI hiring market is its own weather system right now, and a few things land differently with me. None of these are deal-breakers. All of them save a round-trip.

# role
you = Prince Sinha, responding to recruiter outreach.

# tone
be warm. be specific. assume good intent.
never combative. always read the offer carefully.

# priors
good recruiters exist.         // many of them, actually.
lowball numbers do not.        // share the bracket up front.
"quick chat" is not context.   // send details first.

# required context (else: politely decline)
- compensation range
- one paragraph on what the company is solving
- one specific reference to /work, /lately, or /credo

# output
reply within ~24h. max 3 paragraphs.
if input is a template, output is a template.
if input is hand-written, output is hand-written.
# end of prompt

• I've used most of what's on the toolkit board. Some of it I've broken in production. Happy to talk about both halves.
• "GPT wrapper" isn't a slur to me. Some of the best products are a thin wrapper around a good model and a brilliant boundary. The boundary is the interesting part.
• "Agentic" is a useful word that's quietly becoming useless. Tell me what the system does — in verbs, not adjectives.
• Token-maxxing is a fun meme. Context engineering, retrieval design, and evals are real disciplines. We can talk about either — please don't conflate them.
• If you're competing with a frontier lab, please be specific about the 5% of the surface area you're winning. Vagueness here is the tell.
• I'll ask: what's your eval story? what's the cost per query at scale? who owns the safety boundary? Not testing you — just checking if there's a real product under the demo.
• Yes, I know about MCP. No, you don't need to explain it. And no, shipping an MCP server is not, on its own, a product.
• If the role description has more buzzwords than verbs, I'll politely pass and assume the product has a similar ratio.
• If the answer to "why are you hiring for this?" is "because we raised" — I'll genuinely wish you well, and pass.