adi
// distributed, mostly

~ whoami

adi

founder. studied cs in undergrad and grad, specialized in distributed systems at uw-madison, then worked at amazon and google. this page lists a small subset of things i have built over the years.

▸building stained glass

  • distributed systems
  • web3
  • smart contracts
  • nft exchanges
  • ml
  • databases
  • amazon
  • google
  • uw-madison
▤ papers stainedglass.tech ↗

$cat experience.log

  1. founder — stained glassnow

    building stainedglass.tech.

  2. founder — infinity

    a decentralized exchange for nfts. raised $3m in vc funding.

  3. founder — mavrik labs

    pump fun for nfts. incubated at binance labs season 2.

  4. founder — picolo labs

    a decentralized database network for queryable dapp data. raised $2.2m from top silicon valley investors.

  5. engineer — amazon, google

    worked at amazon and google.

  6. grad school — uw-madison

    studied cs in both undergrad and grad. specialized in distributed systems.

$cat backed-by.txt

founded 3 companies in the web3 space and raised vc funding for all 3, including from:

  • binance labs
  • menlo ventures
  • village global
  • ngc
  • nima
  • fj labs
  • sandeep nailwal
  • batuhan dasgin
  • ivan bogatyy
  • ashwin ramachandran
  • + others

$ls projects/ click to expand

› infinity decentralized nft exchange · $3m raised 17

a decentralized exchange for nfts. raised $3m in vc funding.

› mavrik-labs pump fun for nfts · binance labs s2 4

pump fun for nfts. incubated at binance labs season 2.

› picolo-labs decentralized database network · $2.2m raised 7

a decentralized database network to store queryable data. meant to be used by dapps built on blockchains. raised $2.2m in funding from top silicon valley investors.

# demos

# related code

› lstm lstm for predicting crypto prices 1

lstm for predicting crypto prices: current thinking towards methods for training ai algorithms that make price predictions for cryptos and take bets from users are discussed. ai algorithms used here work on two distinct types of data: unstructured in the form of text and structured in the form of exchange data and bets on the platform. while the research on sentiment analysis on unstructured data is vast and several off the shelf solutions are available, we explore an approach based on lstm networks. for structured data, a novel approach is proposed where the state of the platform (all current bets, buzzing topics, randomly timestamped snapshots of past platform state) is reduced to a "game state" and presented as a markov decision process. incentive decay functions that govern the payouts made to users for contributing on the platform are discussed.

› random bots, filesystems, databases, apps 21

$ls papers/*.pdf