I build data infrastructure for extreme-edge and resource-constrained environments. My drive for this started while deploying backend systems for schools in India on "potato" hardware—low-power devices with strictly limited RAM and slow SD cards where standard solutions simply couldn't run. What began as a deployment necessity evolved into a deep personal passion for mechanical sympathy and zero-heap architectures. <p> I recently published my personal Rust primitives—<i>small_collections</i> and <i>bitcraft</i>. These are the foundation for <b>LITSQ</b>, my upcoming zero-heap vector search kernel; they provide the stack-allocated memory and bit-layout control required for the kernel's hot path. <p> Alongside this, I am building <b>SQFLYER</b>, an ecosystem pushing SQLite to its absolute limits. Its foundation is <i>sqlite-ext-core</i>, my thread-safe Rust extension framework. I am currently hardening that core by porting my production-grade materialized views and background job engines to it. <p> The ultimate goal is <b>LITSQ</b>: an O(1) I/O <b>storage-agnostic</b> vector search kernel. It’s a pure compute layer—you can connect it to SQLite, IndexedDB, or S3 to run vector search without loading indexes into RAM. <p> Personal: <a href="https://github.com/dilipvamsi">@dilipvamsi</a> | Upcoming: <a href="https://github.com/sqflyer">@sqflyer</a>