Ricky Moezinia
I am currently building American Terawatt, a national scale power grid for hyperscalers. We build long distance, high voltage, direct current transmission infrastructure to support ai compute growth in the U.S. I learned of the need to scale to GWs of compute while working on Fluidstack prior. I also advise companies on their datacenter strategy and implementation.
Beforehand, I spent many years working in fintech (at a fintech startup Paribus, in quant equities trading, and at Facebook). I had a focus on stablecoins; first building Libra (Diem) at Facebook, and then as the CEO & founder of Alta, a global payments platform built on the first federally chartered bank-backed stablecoin. I was a quant trader in Point72's systematic trading group, Cubist.
I invest through Leo Capital, and on behalf of Sequoia Capital through their scout fund.
I love flying (mostly a Piper Cherokee PA-28), playing poker & padel.
Hide
As a security experiment, I built a module which compromised and harnessed users' CPUs to mine Monero
Liizt
I bought Zola for Latam, Liizt, to test out a thesis around AI transformations in ops-heavy businesses. We're looking for others to run the business now, ping me if interested.
Paribus
I helped build an email scraper to automatically rebate discounts for ecomm purchases and hotel boookings at Paribus (now Capital One)
Prism
A friend and I built a decentralized options exchange on Ethereum in 2015. We created a order execution exchange for ETH options, binding options in Solidity contracts, using Oraclize for price feeds and IPFS to offload data. Apparently this was the second ETH derivatives exchange built globally.
LazyMe
I built a program to do my Spanish Homework for me. Perdón, Señor Garcia! I had a semesters worth (1000s) of online exercises to do; so I wrote a script to navigate through the exercises and submit all the correct answers for me. I got an A.
Lucky 13
Built an autonomous vehicles to race around a ~100m track and to escape an (prior unknown) labyrinth. I worked on localization, mapping, control and path planning.
Puma
I built an iOS application which stores your notes in the cloud encrypted using private keys from the iphone's secure enclave. (Hybrid encryption for top performance and security.) Login with biometrics and a passphrase to decrypt your notes, so noone can see them even if the storage server is hacked.
Triad
I used to build market making and taking bots, using simple graph algorithms to find multi hop cross-exchange price arbitrages. Assets needed to be distributed across exchanges in the right amounts to execute - forecasting arbitrage windows and relocating was more complex than actually finding the mispricings in real-time. With market making back in 2015, I was able to consistently represent spreads representing up to ~10% of prices and for half a year, I was the only market maker for two low volume tokens.
StyleX
Although design is not my forte, I tried to create a personalizable website so you can store all your clothes and belogings online. Not only that, but you can try different outfits on a virtual-YOU to see what looks best. Save the outfits that look good.
Zero Knowledge Range Proofs
We (Novi Crypto Research) published a paper on range proofs. These have loads of applications, from proving income ranges (ZK accreditation) to ZK geolocation.
Paper →Adversarial Infill
Semantic infilling involves filling in significant portions of corrupted images. I explored adversarial methods of achieving semantic infilling on faces. I used a DCGAN (deep convolutional generative adversarial network) trained on facial images to reconstruct portions of images.
Heterogeneous POW Consensus Protocols
I wrote a paper on how heterogeneous network conditions affect distributed systems with a friend at MIT. As blockchain systems scale, and network parameters are taken to the limit, network conditions may impact consensus protocols. I analyzed the impact of heterogeneous network conditions on consensus protocols as block times tend towards their limit. By simulating POW protocols in a heterogeneous peer-to-peer networks, I was able to break consensus by performing an attack with less than 50% malicious hash power. This research aims to raise awareness about disparate network conditions in systems with short interblock times, and suggests alterations to random peering mechanisms such as Bitcoin's.
Digital Currency Initiative - MIT Media Lab
I worked on enabling Discreet Log Contracts in Lightning, a Bitcoin L2. These can primarily be used as onchain BTC futures contracts.
Learn More →