Hey Tron fam,
I’m Bubbly Orca, a data scientist, security researcher, and blockchain developer based in Austin, TX. I made the move from the DMV area to Austin, TX, driven by my passion for crypto, tech, and the rhythm of music. You could say I’m all in on Web3, but I’m equally invested in humanity, believing that the blockchain space holds immense potential for good over the next 10 years, provided we embrace the decentralization aspect of the tech and rediscover our altruistic nature.
The DMV tech scene revolves around government advancement, leading many to boast government experience. As a data scientist with a history in AI/ML and crypto since 2016, navigating the currents of this next wave in crypto has proven challenging. Why the challenge, you ask, when I seemingly have the best of both worlds? Well, as you all know, the crypto space is prone to noise and speculative ventures promising grand returns. Devs bear the brunt of recognition, especially in priming the space.
During bull runs, the focus often shifts to what yields the most profit, overshadowing innovative EIPs or crucial security updates on Metamask to combat phishing. The beauty of the bear market lies in its ability to clear out the noise, providing a conduit for builders to innovate without interruption. I’ve found myself honing my SC development skills, refining methodologies, expanding my vocabulary, and letting my imagination soar. A trend I foresee for builders at the intersection of crypto and AI involves exploring transformers beyond GPT for training.
Transformers like RNN, GRU, LSTM, BERT, ERNIE, or ELMo prove valuable in the CV and NLP domains; GPT isn’t a one-size-fits-all solution. There’s extensive research leveraging well-established transformers in the CV domain for smart contract implications. In my Web3 security domain, smart contracts are condensed into “graphs” or “graph images” called convolutions for smart contract vulnerability detection. Trail of Bits has recently innovated in SC vulnerability detection, using yolov7, a renowned object detection model for real-world objects, to uncover 11 distinct smart contract vulnerabilities.
That’s my hot take for today. If you found this intriguing, follow me on Twitter for more insights. 