I'm Emilie — Data Science Student @ Aix-Marseille School of Economics (AMSE), Rust developer, distributed systems enthusiast, and passionate about language design.
From literature & languages to building programming languages in Rust: curiosity drives innovation.

I build systems that are meant to be read, understood, and taken apart. I work primarily in Rust, across two domains: programming language design and distributed infrastructure — both driven by the same conviction: complexity should be earned, never hidden.

Everything I ship is open source, licensed and built in public.

whispem-lang is a small, self-hosted programming language. The compiler is written in Whispem itself and compiles itself, producing byte-identical output with the reference Rust implementation. It runs on a standalone C VM with no dependencies beyond a C compiler. Rust remains the reference implementation.

Whisper your intent. The machine listens.

• Self-hosted compiler — compiler/wsc.wsp: 1724 lines of Whispem for the full pipeline. Source in, .whbc bytecode out—identical to Rust output.
• Verified bootstrap — The compiler compiles itself. Both outputs share the same SHA-1, ensuring a stable fixed point.
• Standalone C VM — vm/wvm.c: a single-file runtime (~2000 lines) with 34 opcodes, interactive REPL, and a --dump disassembler.
• 204 tests, zero warnings — 153 Rust tests + 51 autonomous C VM tests with bootstrap verification.

fn factorial(n) {
    if n <= 1 { return 1 }
    return n * factorial(n - 1)
}

for n in range(1, 16) {
    if n % 15 == 0 { print "FizzBuzz" }
    else if n % 3 == 0 { print "Fizz" }
    else if n % 5 == 0 { print "Buzz" }
    else { print n }
}

make
./wvm compiler/wsc.whbc examples/hello.wsp   # compile + run
cargo test                                   # 153 Rust tests

dprism — terminal-native data profiling tool in Rust. htop meets pandas-profiling. Explore multi-GB datasets instantly without leaving your terminal. Built with Polars and Ratatui.

minikv — distributed key-value store in Rust. Raft consensus, 2PC transactions, 256 virtual shards, S3-compatible API, 50 000+ writes/sec.

"The best way to learn is to build."

GitHub · LinkedIn · Discord · [email protected]