Founder · Director · Operating Executive
Michael Zeligs, MST
CEO, film director, and Stanford-trained AI research engineer — the rare operator who writes the code, directs the camera, and closes the round. Fifteen years turning early companies into fundable, on-camera stories.
One operator, four disciplines
What Michael actually is
Chief Executive
CEO of ApertureAPI and founder of Start Motion Media. Builds companies from zero through institutional fundraises — and runs the P&L, the roadmap, and the room.
Frontier-AI Trainer
Trains and evaluates frontier models on complex executive-decision work — GDPval-style benchmarks, occupational-complexity evals, and the RLHF harnesses that make the signal reliable.
Producer & Director
Managing producer and director/DP on 2,000+ films — flagship brand campaigns, launch films, and crowdfunding creative that has raised tens of millions.
Image, Audio & Video Specialist
Stanford CCRMA-trained in video engineering and digital signal processing. ffmpeg/AV transcoding, volumetric-video workflows, scoring, and color — the full media pipeline, in one pair of hands.
Image
Audio
Video
“On a shoot, Michael directs. Off it, he writes the Python, designs the funnel, and grades the color. The point of a boutique is that the person who sets the vision is the person who ships it — no hand-offs, no dilution.”
Two ways he gets hired
The same judgment shows up on both sides of the table — whether the client is a founder raising a seed round or a lab training the next model.
For brands & founders
Director & Fractional CMO
Directs flagship brand films and product campaigns, and runs the full launch stack a company needs before it has a CMO — web, video, GTM messaging, customer acquisition, CRO, and the fundraising narrative.
Apple · Amazon · Google · Smithsonian · 100+ founder-led startups.
For AI/ML labs & boards
CXO & AI Research Engineer
Trains and evaluates frontier models on complex executive-decision work, designs the rubrics and harnesses that make that signal reliable, and ships the Python tooling himself.
GDPval-style benchmarking · occupational-complexity evals · RLHF program design.
Frontier AI, in practice
The reason labs hire a real operator to grade models: judgment is the bottleneck, not labeling. Michael encodes senior business expertise into training signal — then builds the tooling that makes it scale.
GDPval
Benchmarks frontier models against real, economically valuable knowledge work — authoring the tasks and rubrics drawn from how executives actually spend their day.
30 hrs → min
Decomposes high-complexity occupational tasks — the kind that take a CEO 30 hours by hand — into structured evaluations a model can be trained and graded against.
45 → 8 min
Built the Python evaluation harness that cut per-task review from 45 to 8 minutes and tripled cohort throughput.
Where the work landed
Not a highlight reel — the public record. Companies Michael built the story and systems for went on to make headlines on their own merits.
$4.75B
Alphabet’s agreement to acquire Intersect Power
Michael led marketing & investor-grade content on the Apple Green Bond–backed Radian Solar build across Intersect Power and Terabase. Deal announced Dec 2025 — Bloomberg, TechCrunch, Forbes.
$130M
Terabase Energy Series C — SoftBank Vision Fund 2
Same solar program, closed March 2025, bringing Terabase past $200M raised. Source: BusinessWire, Mercom Capital.
$34M+
Raised on campaigns he produced & directed
Baubax ($20M), Tap Systems ($13M+), Foldio/Orangemonkie ($1.3M+) — among the most-funded crowdfunding launches in their categories.
Client launches featured in
CNN Money · TODAY · NBC · CBS · Inc. · TechCrunch · Entrepreneur · Huffington Post · Yahoo · Business Insider
Selected work & engagements
Range
The director, the designer, the engineer, and the musician are the same person — which is why the work holds together.
Design
Brand systems, web layout, and campaign art direction — the visual layer that makes a young company look fundable.
Music & sound
Composition and signal processing from Stanford CCRMA — scoring, sound design, and the acoustics research behind two patents.
Engineering
Production Python, C++, and JavaScript daily — automation, data pipelines, and the evaluation tooling behind the AI work.
Patents & publications
- ◆Telecommunications patent — an impedance-matched, mechanically tunable, self-resonant quadrafilar helical phased array for low-power Tx-Rx across OTH radar, UHF near-space, and X-band deep-space regimes.
- ◆Signal-processing patent — a cubic conversion method for bidirectional mapping between electromagnetic and acoustic spectra.
- ◆Published researcher in video engineering and digital signal processing, with Stanford CCRMA collaborations.
Education
Stanford University — CCRMA. B.A. in Media, Science & Technology (MST): video engineering and digital signal processing at the Center for Computer Research in Music and Acoustics. DSP, parallel and real-time systems, filter and codec design, and software synthesis in C++/Python/JS, with foundations in linear algebra, Fourier analysis, and optimization.
Honors
- ◆Diamond Fellowship · Project Horizon
- ◆Prodigy Cohort · Gather Platform
- ◆Stanford StartX · ApertureAPI
- ◆Stanford Creativity in the Arts
- ◆Rotary Youth Leadership Award
Two ways to work together
Book Michael to direct your next launch film — or to lead the marketing, product, or AI-evaluation program behind it.
Start a video project
Hire Michael (CMO / AI)
Petaluma, California · mzeligs@alumni.stanford.edu · Download CV (PDF)