research systems × ecosystem engineering × sharp developer tools
|
Research-oriented systems, built to survive real workflows. I build research systems and evaluation artifacts, then push the same taste for rigor into compiler, packaging, fuzzing, and language-tooling work. Currently working on artifact evaluation in a research lab setting, with a background spanning academic and industry software workflows. |
Current signal agent systems program semantics ecosystem work artifact evaluation Reach acturea@gmail.com WeChat available on request |
research // semantics // toolchains // static analysis // workflow automation
Research · Ecosystem · Tools · Collaboration
Some active work stays intentionally abstract here while under review. The stable through-line is a recurring set of technical problems, not a list of titles.
|
Agentic Software Engineering orchestrated workflows for program transformation, inspection, and constrained automation |
Benchmark & Artifact Design evaluation setups built for reproducibility, diagnosability, and real execution constraints |
|
Program Semantics semantics-aware translation, reconstruction, and transformation rather than surface-only rewriting |
Language-Centric Modeling semantic understanding and structured modeling for language-heavy or annotation-heavy tasks |
I do not treat ecosystem work as a side quest. When the right fix belongs upstream, I would rather land it there than keep a private workaround alive forever. When the ecosystem needs a missing layer, I would rather build it cleanly than leave the gap in place.
|
Nix / nixpkgs compiler packaging, version bring-up, and runtime support work around flang and related toolchain edges
|
LLVM / MLIR / Flang build-system and integration fixes aimed at removing downstream friction instead of papering over it locally |
|
AFL++ low-level fuzzing and runtime-path fixes in systems code where small mistakes quietly become real debugging costs |
Tree-sitter ecosystem grammar authoring, binding maintenance, and parser-upgrade work for domain-specific language tooling |
This is the kind of engineering I enjoy most: work that becomes foundational for other engineers, even when the surface area looks small.
Selected links
Public work tends to be small, sharp, and operational: tools with a narrow surface area and a strong opinion about how the workflow should feel.
|
review-loop durable CLI and daemon for review submission and retrieval, with explicit state, retries, and traceable local artifacts Rust, SQLite |
char-cloud shape-constrained SVG word cloud generator with reproducible output and reusable library APIs Rust, SVG |
|
d2typ data-to-Typst conversion for document workflows that move quickly from structured input to publishable output Rust, Typst |
tree-sitter-paradox tree-sitter grammar and bindings for Paradox scripting languages, designed for better parsing and tooling Tree-sitter, JavaScript, C |
|
scriptmark automated grading tooling for scriptable evaluation workflows with a bias toward practical batch use Python |
modus-foch static analysis tool for Paradox mod playsets, focused on symbol indexing, rule checks, and dependency integrity Rust |
|
Open to research collaboration on systems, tooling, and evaluation benchmark, artifact, and reproducibility work developer tooling, language infrastructure, and static analysis |
Reach acturea@gmail.com WeChat available on request |