Aevo
2021 — 2026 · SG
Chief Operating Officer
Decentralised derivatives exchange backed by Paradigm, Dragonfly, and Coinbase.
Gabriel Ravindran
Singapore
Operator — factories, exchanges, consumer manufacturing.
I spent five years as COO of Aevo, a decentralised derivatives exchange with more than $100B in lifetime trading volume.
Before that I sold defect-detection AI into semiconductor and solar fabs, and built consumer-manufacturing businesses in Malaysia.
I’ve invested in 20+ startups. These days I’m thinking mostly about robotics, industrial AI, and AI infrastructure.
Work
FireVisor
2019 — 2021 · SG
Head of Business Development
Real-time defect-detection AI for manufacturers. Sold into semiconductor and solar fabs on five continents — from the first meeting to the system running live on the line.
Allure Laboratories
2017 — now · KL
Co-founder & Director
Contract manufacturing of personal-care and household goods for some of Asia’s largest organic retailers.
Bacoff
2013 — 2016 · KL
Co-founder
One of Asia’s first organic household-care brands. Scaled to 2,000 stores across the region.
Some things I believe
The AI bottleneck is never where the headlines point.
- Chips had their turn; power is having its. The real queue is downstream — the Korean transformer makers are sold out for years.
- In optics it’s two layers down: substrates and epitaxy, a handful of companies most people have never heard of.
Humanoid robots are a supply-chain problem, not an AI problem.
- China gates them twice: the magnets, and the machines that make the magnets.
- The demo is marketing. The deployment is the business.
The best work has the slowest scoreboard.
- Give a team a live score and the score becomes the roadmap.
Software isn’t cheap. The code is.
- When everyone can build the product, buyers stop comparing products. They pick who they trust in prod.
- The code takes a weekend. The trust takes years.
Every “organic” market was paid for.
- Market makers wanted rebates. Supermarkets wanted margins. Same negotiation.
- Most growth problems are incentive problems.
The obvious thing is usually still available.
- “Someone must be doing it” is how it stays undone.
- Every field is smaller than its conference suggests.
Atoms don’t have version control.
- A bad release rolls back in minutes. A bad batch is already on a truck.
- What software people call slow is what no-undo looks like.
What someone celebrates is what they’ll optimize.
- A resume shows what someone did, never what they were chasing.
- Validation is a hit. Progress is an income.
Contact
If you’re building in robotics, industrial AI, or AI infrastructure, I’d like to talk — as an operator or an investor.