Marco Sanvido: background check and fit as an AI-startup co-founder and CEO

Prepared 16 September 2026 from public web sources only (20 sites visited; list at the end). The request named "Marco Sandivo"; that spelling returns zero results on Google, which auto-corrects to Marco Sanvido. Every source below (ETH Zurich thesis, UC Berkeley postdoc, Belmont CA, MemVerge, Robust Intelligence) points to one person, Marco Aurelio Antonio Sanvido. If you meant someone else, stop here.

Bottom line

Who he is. A 20-plus-year Silicon Valley infrastructure engineer and engineering executive: ETH Zurich PhD (2002), UC Berkeley postdoc with Tom Henzinger, VMware and Hitachi researcher, early engineer then Director of Engineering at Pure Storage (about 7 years, through the IPO), VP Engineering at three venture-backed startups in a row (Styra, MemVerge, Robust Intelligence), a 7-month incubation stint at Sutter Hill Ventures, then a Visiting Partner at South Park Commons. 59 granted US patents, h-index 24. No adverse media, litigation or controversy found.

As a co-founder: strong. He has built and scaled 0-to-50-engineer organizations four times, has taken a first product to GA, has run enterprise-readiness and compliance programs that preceded a roughly $400M acquisition, is hands-on with LLM agents today (718 GitHub contributions in the last year), and writes clearly about moats and startup strategy. He is available: he left SPC, describes himself as an independent advisor and angel, and tagged a September 2026 prototype "Interview".

As CEO: unproven, and his own words point the other way. Every senior role has been the "how" person next to founder-CEOs; he calls his style "servant leadership", says "the goal is not to have a fantastic exit", and in March 2026 described "my own transition into advisory work". There is no public evidence of him selling, raising capital, or owning P&L. The natural fit is co-founder and CTO/COO alongside a CEO who sells and raises. If he must be CEO, the diligence below tells you exactly what to test.

1. Verified career timeline

Dates before 2011 are approximate (press bios only). Startup dates come from LinkedIn tenure snippets shown in Google results (the profile itself is behind a login wall): Pure Storage 7 y 1 m ending about Nov 2017; Styra Nov 2017 to Aug 2020; MemVerge Aug 2020 to Sep 2021; Robust Intelligence 2 y 10 m to about Jul 2024; Sutter Hill Ventures Aug 2024 to Feb 2025; South Park Commons from Mar 2025.

Period	Role	What the record shows
to 2002	ETH Zurich, MS and PhD in CS (Computer Systems Institute, the Wirth/Gutknecht Oberon group)	Thesis "Hardware-in-the-loop simulation framework" (2002). Built the software core of ETH's autonomous model-helicopter autopilot; co-authored the IFAC paper on HIL testing with temporal logic. Says on a 2023 podcast: "we did a startup there", his first startup and, per a 2024 post, the one he co-founded. Name and outcome not found.
2002 to 2004/05	Postdoc, UC Berkeley EECS (Tom Henzinger's group)	Giotto real-time programming (IEEE Control Systems Magazine 2003, 215 citations), Extreme model checking (2003, 130), logical execution time (HSCC 2004), VEE 2005 microkernel. Rigorous formal-methods and embedded-systems pedigree.
about 2005 to 2010	Researcher, VMware; then Hitachi Global Storage Research	VMware core hypervisor (binary translation). At Hitachi: flash, SSD and shingled-disk work; 15 patents; "Flash devices with RAID" patent has 269 citations; Proceedings of the IEEE paper on NAND flash (2008).
about 2010 to Nov 2017	Pure Storage: early engineer, later Director of Engineering	Joined pre-IPO, stayed two years past the Oct 2015 IPO. 35 patents (RAID parity, encryption, dedup). Led a "stealth project". Listed as Pure's contact on its FIPS 140-2 crypto validation (2017). Acknowledged in the SIGMOD 2015 Purity architecture paper. He calls Pure's engineers "some of the best engineers I ever worked with".
Nov 2017 to Aug 2020	VP Engineering, Styra (Open Policy Agent company)	"Helped bring their first product to GA" (MemVerge press release). 8 patents on API authorization and permission analysis with the Styra founders.
Aug 2020 to Sep 2021	VP Engineering, MemVerge	Hired to lead the Memory Machine (persistent-memory) team under CEO Charles Fan. Left after 14 months, about ten months before Intel discontinued Optane, the product's hardware base. Shortest tenure; worth asking about.
about Oct 2021 to Jul 2024	VP Engineering, Robust Intelligence (AI security, 51 to 200 staff)	Ran a Kubernetes/Go platform with a Python ML engine deployed in customer VPCs; led the SOC 2 Type 2 program that passed with zero deviations; company named an SF Chronicle Top Workplace. Cisco announced the acquisition (reported at about $400M) a few weeks after he left; he congratulated the team from outside.
Aug 2024 to Feb 2025	Sutter Hill Ventures (incubations)	7 months in Palo Alto; SHV is the firm that incubated Pure Storage and Snowflake. Nothing was announced; either an incubation that did not proceed or an EIR period.
Mar 2025 to about late 2025	Visiting Partner, then Technical Advisor, South Park Commons	Announced Mar 2025. Not on SPC's team page as of Sep 2026. Publicly congratulated SPC-backed Nuance AI on its $50M Series A (Sep 2026).
2025 to now	Independent: "Eng(el): advising startups engineering efforts", angel	Writes monthly LinkedIn essays; ships small AI-agent prototypes on GitHub almost weekly; built an interview take-home ("protostrata") in three days in Sep 2026, so he is actively looking at his next role.

2. Evidence-based strengths

Repeated 0-to-50 engineering scale-ups. Four venture-backed startups as VP or director; his own handbook ("Engineering Management: The First Dirty Dozen", revised Apr 2026) is explicitly about 0-to-50-engineer orgs and is written from practice, not theory.
Enterprise infrastructure depth that is hard to fake. Storage, encryption, RAID, hypervisors, Kubernetes SaaS, policy/authorization, and AI-security platforms. 59 granted patents with 682 forward citations; h-index 24 across academia and industry.
Enterprise-readiness muscle. FIPS 140-2 at Pure, SOC 2 Type 2 at Robust Intelligence, customer-VPC data-plane deployments for regulated industries. This is exactly the work that makes an AI startup sellable to enterprises and acquirers.
Current, hands-on with AI agents. 718 GitHub contributions in the last 12 months across Python, Go and TypeScript: agent marketplaces, Claude Code skills, a Go port of nanochat, an eval-driven regulatory-intelligence workspace with GEPA prompt optimization. Self-taught path is visible in forks (nanoGPT 2023, LLMs-from-scratch 2024, reasoning-from-scratch 2025).
Judgment about hype. Nov 2023: LLMs are "fantastic to demo" but business-critical use needs guardrails; "if everybody has it, nobody has it" about AI-generated core IP; warns of 2000-style over-investment. Sep 2026: predicts personal agents get absorbed into operating systems and distribution owners win. These are the views of someone who has watched three platform transitions (virtualization, flash, Kubernetes) from the inside.
Founder-evaluation reps. At Sutter Hill and SPC he spent about 18 months "mentoring, advising, and evaluating many ambitious founders" and wrote a usable filter: ask about the "how", not just what/why/why-now.
People leadership with a track record of retention. Servant-leadership stance, measures impact rather than commit counts, adapts to life phases; Robust Intelligence was an SF Chronicle Top Workplace under his engineering leadership.
Clean record. No lawsuits, terminations, fraud or controversy surfaced in adverse-media searches. Long tenures where it mattered (7 years at Pure).
Bay Area, well networked. Belmont, CA. Network spans Pure/SHV alumni, Styra/OPA, Robust Intelligence/Cisco, SPC founders, and Berkeley EECS (Henzinger group).

3. Weaknesses and risks for a CEO seat

No evidence of selling or raising. Twenty-plus years of public record and not one reference to closing a customer, running a sales team, pitching investors, or owning revenue. His own framing: "in a startup, the biggest gap is always the one who helps you close the customer" (said about others). A CEO's first two jobs are recruiting capital and customers.
Self-positioned as the builder behind founders, not the visionary. "The companies I was fortunate to join"; "the founders had the fantastic vision"; "the car around the engine". He has been the number-two technical executive four times and has never been the final decision-maker at a US company.
Stated direction is advisory. Mar 2026: "navigate my own transition into advisory work". Current headline: "Advisor, ex-VP of Engineering, Angel". Sep 2026 GitHub tag "Interview" suggests he is weighing employment, not only founding. Motivation for a full CEO commitment must be tested directly.
Explicitly not exit-driven. "The goal is not to have a fantastic exit... it's a lottery ticket... it's the journey." Admirable in an engineering leader; investors backing a CEO usually want to hear the opposite.
Short MemVerge tenure (14 months). Possibly a market call (persistent memory was about to lose its only supplier), possibly a fit problem. Ask.
Left Robust Intelligence weeks before the Cisco deal. Neutral to positive if it was planned (he congratulated the team publicly), but confirm he was not pushed out and understand his equity outcome, since it affects his personal runway and hunger.
Limited public voice. About 3.8K LinkedIn followers, 401 on X; essays get 20 to 40 reactions. Founder-CEOs in AI today recruit and sell partly through public presence. He would need a co-founder or hire who does that.
Not an ML researcher. Zero AI/ML publications; AI depth is that of a serious systems engineer who has spent three years building around models and using agents daily. Fine for an applied-AI infrastructure company; a liability if the startup's edge is model research.
Career stage. PhD in 2002 puts him around 50. This brings judgment, network and patience, and less tolerance for a founder's salary and 80-hour weeks; verify what he actually wants.
Solo prototypes have no users. All 48 GitHub repos have zero stars. He can build fast; nothing shows he has taken a product from zero to paying users on his own.

4. CEO-fit scorecard

Bar length reflects the weight of public evidence found, not a prediction of ability. Red rows are where public evidence is absent, which is different from evidence of weakness; they are the rows to probe in conversation and reference calls.

5. Assessment

As a co-founder

The case is strong. He brings what most AI-startup founding teams lack: someone who has actually built the reliable, compliant, deployable product around a clever engine several times, who can hire and keep engineers, and who understands what enterprise buyers and acquirers check. He is technically current in the specific area that matters right now (agents, evals, harnesses) and can prototype end-to-end alone. His stated philosophy, take a proper foundation first so you can accelerate later, and build a "how" that is hard to copy, is the right instinct for a technical co-founder. He is also, by every public signal, decent to work with.

As CEO

Unproven rather than disproven. Nothing in the record shows him doing the three things a startup CEO cannot delegate: setting and selling the vision to customers, raising money, and being the person who is ultimately accountable. He has consistently described himself as the person who executes a founder's vision, and as recently as March 2026 as someone moving into advisory work. Competence in a CEO role would depend on him wanting it badly enough to change what he has optimized for over two decades, and on pairing with someone who covers distribution.

Two configurations make sense:

Sanvido as co-founder and CTO (or President/COO), with a commercially-minded CEO. This uses everything he has proven and none of what he has not.
Sanvido as CEO only if the company is an infrastructure or developer-tools product sold to engineers (where a credible technical leader can sell early), the first commercial hire is senior, and he says convincingly, unprompted, that he wants to run the company rather than advise it.

6. Questions to ask him and reference calls to make

Ask him directly

Why did you leave MemVerge after 14 months, and Robust Intelligence just before the Cisco announcement? What was your equity outcome, and how much personal runway does that give you?
What happened in the seven months at Sutter Hill Ventures? Was there an incubation, and why did it not proceed?
Tell me about the startup you co-founded at ETH: what it did, how it ended, what you learned.
Describe the largest customer deal you personally drove to close, and the last time you pitched investors.
In March you wrote about transitioning into advisory work and you are interviewing now. What would make you commit full-time to founding, for five to seven years, as the accountable leader?
You wrote "the goal is not to have a fantastic exit". How do you reconcile that with taking venture money?
How would you split CEO/CTO responsibilities with me, concretely, in the first 12 months?

Reference checks (public co-workers, all findable)

Charles Fan, MemVerge CEO, who hired him and saw him leave after 14 months.
Tim Hinrichs and Teemu Koponen, Styra co-founders and his co-inventors, on taking a first product to GA.
Yaron Singer and Kojin Oshiba, Robust Intelligence founders, on three years running their engineering org and the exit.
John Colgrove and Ethan Miller at Pure Storage, on seven years of technical contribution and the stealth project.
Tom Henzinger or Christoph Kirsch (academic), if you want a read on how he was as a researcher.
Engineers who reported to him at Robust Intelligence or Styra: does the servant-leadership description hold, and did they follow him between companies?

7. What this check did not cover

LinkedIn is behind a login wall; tenure dates come from Google result snippets of his profile and could be slightly off.
No criminal, credit, court-records, education-verification or employment-verification checks were run. A commercial background-screening service should do those before any equity agreement.
Crunchbase, Bloomberg, RocketReach, ResearchGate and DBLP blocked automated access; their data was captured only via search snippets.
The ETH-era startup he co-founded could not be identified.

Sources visited

Google search "Marco Sandivo" (zero exact results; auto-corrects to Sanvido).
South Park Commons blog, "Marco Sanvido Joins SPC as Visiting Partner", 6 Mar 2025: https://blog.southparkcommons.com/p/marco-sanvido-joins-spc-as-visiting-partner
Google Scholar profile: https://scholar.google.com/citations?user=9H8iPS8AAAAJ
Google search results with LinkedIn tenure snippets (MemVerge, Styra, Pure Storage, Robust Intelligence, Sutter Hill Ventures).
PR Newswire, "Marco Sanvido Joins MemVerge as Vice President of Engineering", 25 Aug 2020: https://www.prnewswire.com/news-releases/marco-sanvido-joins-memverge-as-vice-president-of-engineering-301117403.html
YSecurity case study on Robust Intelligence SOC 2 and Cisco acquisition, 22 Sep 2024: https://ysecurity.io/case-studies/soc-2-type-2-robust-intelligence/
GitHub profile and repository list: https://github.com/msanvido
Engineering Management handbook: https://msanvido.github.io/eng-managers-handbook/
Google search for ETH Zurich thesis and helicopter project (ETH Research Collection, Springer, ScienceDirect snippets).
Tractable podcast transcript, "Robust Intelligence VP Marco Sanvido: Securing LLMs", 15 Nov 2023: https://tractable.withorb.com/episodes/robust-intelligence-vp-marco-sanvido-securing-llms/transcript
LinkedIn article "The Ambition Is in the How", 4 Feb 2026: https://www.linkedin.com/pulse/ambition-how-marco-sanvido-cgsvc
LinkedIn article "Moats in the Age of AI", 11 May 2025: https://www.linkedin.com/pulse/moats-age-ai-marco-sanvido-ck1ic
LinkedIn article "The World of AI Is Moving Fast: Are You Too?", 30 Mar 2026: https://www.linkedin.com/pulse/world-ai-moving-fastbut-you-marco-sanvido-anhbc
Adverse-media Google search (lawsuit, fired, fraud, complaint, controversy, court): no relevant results.
IDiyas USPTO inventor summary: https://idiyas.com/inventor/marco-sanvido
X profile: https://x.com/msanvido
Google search for Sutter Hill Ventures, Pure Storage history (Purity SIGMOD 2015 paper, NIST FIPS 140-2 listing, Slideshare bio, Google Patents).
South Park Commons team page, Sep 2026: https://www.southparkcommons.com/team
GitHub repository protostrata: https://github.com/msanvido/protostrata
LinkedIn article "The Founder's Mode: A Personal Perspective", 17 Sep 2024: https://www.linkedin.com/pulse/founders-mode-personal-perspective-marco-sanvido-g9r6c

Blocked or unavailable: LinkedIn profile page (auth wall), Bloomberg, Crunchbase, RocketReach, ResearchGate, DBLP, ETH Research Collection search.