I build AI products end to end, from the agent runtime to the cloud it runs on. Engineer at TMLS.NYC, a New York venture studio.
ronmuj2808@gmail.comI have worked with Claude Code and OpenAI Codex since their launch, and they are at the core of how I build.
Multi-agent workflows, custom skills, hooks, and MCP integrations. Agents carry the volume; I own the architecture, the review, and the result.
On the model side: prompt and context engineering, retrieval and reranking, evals and LLM-as-judge, local and on-device models, and early hands-on work with new model types like Jev by TypeSafe for fast, calibrated decisions.
Ten AI products on one shared agent platform, from multi-agent simulation and competitive intelligence to an assistant that runs fully offline on the phone.
I build the agent runtime and the products on top of it, plus the shared login, design system, and release pipeline behind all of them.
A home for creators and their communities: livestreams with chat and replays, scheduled shows, creator payouts, and automated content moderation.
I own the cloud infrastructure and the delivery pipeline, so every change goes from code to production tested and reproducible.
Comparisons and ratings of crypto exchanges, wallets, brokers, and more, across eight product categories.
I replaced the editorial team's spreadsheets with a full-stack dashboard. One configurable scoring engine rates every product live while editors type.
A blockchain built around native account abstraction, so wallets behave like normal accounts instead of seed phrases.
I added passkey signing (WebAuthn, P256) to the chain's Ethereum client and brought it into the ERC-4337 bundler.
German, Albanian, English