When I founded inDinero in 2009 — at 19, with my classmate Andy Su — I tried hiring engineers from everywhere. Bay Area. Eastern Europe. India. We made every mistake in the book.
By the time I stepped away from the CEO seat in 2022 and started Mahway — a venture creation firm now backing a $1.5B combined portfolio — I had thirteen years of pattern-matching across what worked and what didn't in distributed engineering.
In 2023, the math tilted in a way I hadn't seen before. This post is why FutureProofing exists.
The talent math broke around 2023
Three things happened in parallel:
-
US senior AI engineer comp went vertical. The Levels.fyi cohort for "Senior AI/ML Engineer" jumped from $260K total comp in 2021 to $370K+ by 2024. With recruiter fees, equity dilution, employer tax, and benefits, the loaded monthly cost cleared $24K — and that was before sign-on premiums for hands-on LLM experience.
-
The hiring timeline didn't shrink. It got longer. From job post to first PR in production: 6+ months for a senior AI hire. Three of those months were just sourcing, two were interviews, one was onboarding. The market priced AI engineers like 2021 software engineers but moved at the speed of 2018.
-
AI tooling made geography matter less. Cursor, Claude Code, and Copilot collapsed the gap between "elite engineer" and "good engineer with great tools." Suddenly a senior AI engineer in Bogotá using Claude Code Max was shipping at the same velocity as a Bay Area peer using Cursor — at 40-60% lower loaded cost.
The first two trends made US in-house hiring untenable for most non-FAANG companies. The third made LATAM specifically — not Eastern Europe, not South Asia — the right answer.
Why LATAM specifically
I'm asked this constantly. "Why not Poland? Why not the Philippines? Why not just hire from anywhere?" Three reasons.
| Factor | LATAM | Eastern Europe | South / Southeast Asia |
|---|---|---|---|
| Timezone overlap with US | Full PT/ET overlap | 5–8 hours offset | 9–12 hours offset |
| English fluency | High among senior engineers | High | Mixed |
| Same-day code review | Yes | Practically no | No |
| Cultural distance from US tech | Low (deep Bay Area diaspora) | Moderate | Higher |
| Senior AI talent depth | Growing fast (Rappi, Nubank, MercadoLibre alumni) | Mature | Mature, but volume diluted |
The killer feature isn't price. It's same-day code review and synchronous pair programming. When your engineer in São Paulo opens a PR at 10am their time, your US team reviews it before lunch. When your engineer in Kraków opens one, you get it next morning.
Compounded over a year of sprint cycles, that's the difference between a 3-week feature ship and a 6-week one. LATAM gives you the cost arbitrage without the velocity tax.
What I learned at Mahway about embedded vs FTE
Mahway invests in three startups a year. We're selective by design — we'd rather pick three founders we'll back to a $100M outcome than spray and pray across thirty.
That selectivity rubbed off on how I think about engineer placement. The moment a portfolio company hits Series A and starts hiring AI engineers, the same pattern recurs:
- Months 1–6: 1–2 unfilled senior AI roles. Burn rate keeps ticking. Founders hire recruiters, run 8 weeks of interviews, finally close someone — who turns out to be junior dressed in senior salary.
- Months 7–12: realization that the hire is below bar. Cut losses. Repeat the cycle. Now you've burned a year and 12 months of runway.
The embedded model — not contractor, not outsourcer — fixes this. Engineers live in your codebase, your Linear, your standups. They're not arms-length staff aug. They're functionally part of your team for as long as you need them.
The math from our portfolio:
- Traditional FTE path: $288K loaded for 12 months, 6 months ramp, 6 months productive output.
- Embedded path (FutureProofing): $162K all-in for 12 months, 3 weeks ramp, 11.5 months productive output.
That's $126K saved and 5 months of opportunity cost recovered. For a Series A startup, that's the difference between hitting Series B metrics and not.
The FutureProofing thesis: 12 of 2,000
The hardest part isn't sourcing. LATAM has plenty of senior AI engineers. The hard part is filtering for the ones who can ship production AI systems — not just demo notebooks.
We contact roughly 2,000 senior AI engineers each month. We accept 12. The 99.4% rejection isn't about cruelty — it's because most "senior AI engineers" in 2026 have shipped one fine-tuning experiment and put "AI engineer" on LinkedIn. Real production AI engineers — people who've debugged a RAG pipeline at 11pm, written eval harnesses, deployed multi-agent systems with observability — are 0.6% of the pool.
Our 5-stage funnel (detailed here):
- Initial screen for shipped systems with real users
- Technical assessment — bug-hunting on a real production code snippet
- EQ + behavioral — how they communicate ambiguity
- Paired AI challenge — live work in Cursor or Claude Code
- Final interview — judgment under pressure
If they can't survive all five, they don't go on the bench. As of today's update, we have 3 engineers immediately available, 5 ramping next month, and 14 total in the active bench, growing 3+/month.
What I'd tell a founder reading this
I've sat on both sides of this. As a founder, I bled on hiring mistakes for years. As an investor, I watch portfolio companies repeat my mistakes every quarter.
If you're a Series A–B founder reading this and you've been told the path is "post a job, wait 6 months, hope": there's a faster path. We've already done the vetting. The 12 of 2,000 are on the bench. You can have one embedded in your team in 3 weeks, at 40-60% the loaded cost of a US in-house hire, with no equity dilution.
The companies that figured this out in 2024 are the ones quietly outshipping their peers in 2026. The math doesn't care about preferences — it just compounds.
If this resonates, book a 30-minute call. No commitment. We send 2-3 vetted profiles in 48 hours, and you decide whether the model fits.
— Jess
FAQ
Why is Jess Mah betting on LATAM AI engineers instead of US or Eastern European talent?
The math broke around 2023. A US senior AI engineer costs $24K+/mo loaded and takes 6 months from job post to first commit. A senior AI engineer in São Paulo, Bogotá, or Mexico City has the same English fluency, the same access to training, and full PT/ET overlap — at a fraction of the cost. Eastern Europe was the previous arbitrage but the time zone gap (5–8 hours) breaks pair programming and same-day code review. LATAM doesn't.
What's Jess Mah's background and why does it apply to AI hiring?
Jess founded inDinero at 19 — scaled it to 150 employees and a nine-figure valuation. She's now Executive Chair of Mahway, a venture creation firm with a $1.5B combined portfolio across fintech, biotech, AI, and legal tech. The pattern she's optimizing for is capital-efficient hiring — the same lens that built inDinero applied to the AI talent market.
What does FutureProofing's bench size look like vs traditional staffing?
Currently 14 senior AI engineers in the active bench, growing 3+/month, with 3 immediately available and 5 ramping next month. The funnel is 12 accepted out of 2,000 contacted monthly — roughly 0.6%. Traditional staffing platforms screen for breadth across thousands of verticals; we screen for depth in production AI systems specifically.