My technical foundation began in cybersecurity and cryptography. As a teenager, I taught myself penetration testing, ethical hacking, and system exploitation. That early work gave me a deep understanding of encryption, anonymity, and decentralized systems, and how they behave under real-world pressure.
By 2014, I was active in Bitcoin, Monero, and cypherpunk communities, experimenting with mining, privacy-preserving protocols, and peer-to-peer financial systems. These experiences shaped how I think about blockchain, from its cryptographic primitives to its practical applications in global finance.
Blockchain & trading infrastructure
At Avo, I architected a multi-chain trading system that spans Discord, Telegram, Solana, and Ethereum. This includes:
- Wallet orchestration systems with encrypted key storage and pooled execution.
- Copy-trade infrastructure that replicates whale and agent trades in real time across user wallets.
- A Jupiter-based DEX aggregator, and an aggregator-of-aggregators that minimizes gas and slippage through pre-trade simulation.
- A Solana transaction parser and a Geyser gRPC stream transformer for high-throughput event processing and on-chain liquidity pricing.
- Structured signature-based security models for zero-click copy-trading that preserve UX simplicity and cryptographic integrity.
Systems architecture
I've engineered distributed backends combining Supabase (Postgres + Functions), ClickHouse for analytics, etcd and Redis for rapid key-value access, and NATS JetStream for real-time event streaming. These systems power wallet tracking, swap execution, trading agent performance, and live analytics pipelines at scale.
My infrastructure experience spans Kubernetes, Docker, AWS, Railway, and Vercel. I've deployed systems optimized for high throughput, cost efficiency, and seamless user experience across web and mobile.
Machine learning & AI
At Vocally, I led the development of proprietary voice AI and NLP models trained on over 1.5M real-world phone transcripts. I built supervised learning pipelines for sentiment analysis, intent detection, and diarization, optimized for noisy telephony environments, including real-time transcription, keyword spotting, and batch inference APIs built with Python and TensorFlow.
AI agents & LLM tooling
I build at the application and infrastructure layer of agentic systems. With Agent Grid, I designed an infinite-canvas workspace that orchestrates multiple Claude Code instances, terminals, browsers, and notes inside a single environment, with role-based workers, MCP integrations, and cross-agent messaging.
I also research persistent context: TAG Memory is a topic-scoped vector memory architecture I built and wrote up to address context pollution in long-running LLM conversations, paired with an open-source MVP. Earlier work includes agentic-ui, an experiment in LLM-controlled surface architectures where the agent composes layout and components, and betternotes, a focused take on AI-native note-taking.
Product engineering
Across ventures, I've owned full-stack development using TypeScript, Node.js, Next.js, React Native, and Go. I've integrated payments (Stripe), authentication (WalletConnect, Dynamic, Privy, and signature-based), and analytics (Mixpanel, PostHog) into consumer apps that reach thousands of users. My approach blends systems design with user experience, making advanced tech intuitive and usable.
In short, my background spans security, AI, blockchain, and large-scale system design. Every product I've built connects those disciplines to create software that is secure, scalable, and designed for real-world adoption.
