Full-Stack & Systems Engineer

Building reliable
software from
kernel to cloud.

Open-source contributor focused on performance, correctness, and fixing bugs that matter — from 21-year-old rsync edge cases to production SQLAlchemy crash chains.

$
a3f9b2cfix: preserve empty dirs under --backup-dir (rsync)71e0c4afeat: --report-format json versioned schema v1.09d2b81ffix: SQLAlchemy lazy-load crash on /test/ pagescc4f1a8fix: false-positive CI status on 0 comparison rows

Projects

SyncIt

SyncIt

Sync your liked songs and playlists across Spotify and YouTube Music — seamlessly.

Next.jsNext.jsTypeScriptTypeScriptEExpress.jsPostgreSQLPostgreSQLRedisRedis
WildGuard

WildGuard

AI-powered wildlife rescue platform connecting people with the right NGOs instantly.

ReactReactTypeScriptTypeScriptPostgreSQLPostgreSQLFirebaseFirebaseDockerDocker

recon-oss

Daily LLM digest of merged PRs and open issues, personalized to your OSS contribution goals.

TypeScriptTypeScriptNode.jsNode.jsSQLiteSQLiteOOllamaAAnthropic

Selected Contributions

0yr
rsync bug resolved
0K+
HTTP 500s fixed
0
media files tested
0
upstream repos
Infrastructure · rsync

Fixed a 21-Year-Old rsync Bug

--backup --backup-dir --delete silently dropped empty directories. Files were backed up correctly but the directories themselves were deleted rather than preserved in --backup-dir. Root cause: delete_item() routed S_ISDIR directly to do_rmdir_at(), bypassing the backup-dir logic entirely. Fix uses rmdir-first as atomic emptiness proof, then mkdir in --backup-dir. Also handles Solaris EEXIST quirk.

CSystemsrsync#967Solaris compat
delete.c
/* old: S_ISDIR → rmdir, no backup */ do_rmdir_at(dfd, fname); /* fix: rmdir proves directory empty */ ret = do_rmdir_at(dfd, fname); if (ret == 0 && backup_dir) make_bak_dir(backup_dir_buf); /* Solaris: EEXIST == ENOTEMPTY */ if (errno == EEXIST || errno == ENOTEMPTY) return DR_NOT_EMPTY;

Philosophy

Engineering as
accountability.

I work closest to the problem. That means reading stack traces in production logs, querying databases for actual evidence before proposing fixes, and tracing bugs to their root — not just their symptoms.

Most of my contributions exist because I found something broken and shipped a fix with proof. A 21-year-old rsync edge case. Three thousand HTTP 500s that nobody had connected into a single root cause. A false-positive CI status that had been quietly lying for years.

I move between layers — C internals, Python backend, CI pipelines, packaging — because reliability problems rarely stay on one level.

01
Evidence before hypotheses

Reproduce on production data. Run the query. Read the log. Then explain what's happening.

02
Backward compatibility is a contract

The rsync backup fix adds new behavior without changing existing behavior. That's not a compromise — it's precision.

03
Upstream first

Fixes that only help one codebase are half-finished. Real fixes go upstream with reproducers, test cases, and rationale.

04
Systems depth enables product breadth

Understanding how rsync traverses file trees makes me a better backend engineer. Low-level knowledge compounds.

Technical Range

Systems
C · POSIX systems
Rust
Memory & linking
rsync internals
Cross-platform build
Backend
Python · SQLAlchemy
TypeScript · Node.js
REST API design
DB forensics / SQL
C# · .NET
Infrastructure
GitHub Actions
Snap packaging
CMake · Make
Docker
Linux / Ubuntu
Debugging
Root cause analysis
Production log triage
Regression testing
CI reliability
Media codecs · CC

Contact

Let's build
something solid.

Open to internships, GSoC collaboration, and interesting open-source work. If you've found a hard bug and need someone who reads the source code, reach out.