Skip to content
View Xaena53's full-sized avatar

Block or report Xaena53

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
Xaena53/README.md
Xaena53 — Full-Stack Developer: backends, computer vision and AI tooling currently building Warden of the Fields

Activity — the grid first signs BEDİRHAN ÇETİN, then turns into the contribution heatmap of the past year, with weekly commits per project

Asfalya — vehicle-mounted road damage detection for municipalities

Computer vision for municipal road maintenance. Phone cameras ride on city buses; a YOLO11 detector trained on 35,984 images finds potholes, cracks and sunken manholes; GPS interpolation and OSM map-matching pin each one to a street, and duplicate sightings across passes collapse into a single tracked repair. Faces and plates are blurred before detection, and the pipeline refuses to process an unmasked frame at all, so the privacy rule cannot be forgotten. Node/Express + PostGIS behind a MapLibre panel, with weekly PDF reports for the public works desk. Presented as my internship project at BAUM, Anadolu University.


Asfalya screens — damage map of Eskişehir and two detection frames

Aegis — network-verified approvals for AI agents that spend money

Aegis hands Google Ads and Meta to a language model and keeps the budget in human hands. Approval happens on the server: when the client supports MCP elicitation the server asks the person directly and ignores the agent's own confirm flag. Before that question is even shown, the mobile network is asked through GSMA Open Gateway / CAMARA whether the approver's SIM was swapped recently, and a failed or silent network check refuses the action. Campaigns are created paused, the account owner sets a ceiling the model cannot raise, and scraped site copy is treated as untrusted input. Started as AdsPilot; open source under AGPL-3.0.


AnimeRank media pipeline architecture

A video platform for anime that I designed and built from the interface to the servers. Sources arrive over P2P, pass a season guard and a quality score, and are re-encoded to H.264 that every browser can play; selected series get a 4K pass through an Adore upscaling model running on TensorRT. Around it: MyAnimeList and AniList two-way sync per profile, multiple profiles per account, a community layer with daily quests and levels, a subtitle checker that measures timing against speech in the audio, and deployments that swap between two instances without dropping a request.


Kurenai subtitle translation pipeline

Batch translation for anime subtitles. Karaoke, song and on-screen effect lines are filtered out before anything is sent, which also cut the cost of a typical episode several times over. A line-shift gate rejects any episode whose translated lines drifted out of step with the source, and nothing reaches viewers until it has passed the quality gates. A budget cap governs the translate step.


Kurenai screens — kurenai.shop and a station sign redrawn in Turkish

RepChat live-support architecture

A live-support runtime running in production: embeddable widget, WebSocket gateway, category-routed queues, and an AI greeter that resolves the easy tickets before a human ever sees them.


RepChat screens — operator inbox and the phone widget

Production hardening — defense in depth

More work — EKT Akademi, Warden of the Fields, Demir Cephe, Hapivanak, Çakmak Pen, MHRS data module

Warden of the Fields — Unreal Engine 5.8 captures of the village

Client sites in production — EKT Akademi and Çakmak Pen

Capability tree — 33 technologies grouped by role, each branch naming the projects that use it

Grouped by the job each tool does, and every branch names the projects it is used in. The list was read off the dependency manifests in those repositories, so it records what ships.


Toolchain orbit — the same 33 technologies on three rotating rings

Live build — the Aegis approval gate typing itself out

Engineering core competencies

Verified metrics — 2,720 passing tests across six projects and the road-damage model training curve

Every number on this page is measured: test counts come from the suites that gate each push, and the training curve is the actual run that produced the road-damage model, including the class it is still weak at. A passing test only proves a line ran, so the Aegis suite is held to a stricter standard: each guard is broken on purpose to confirm a test notices.


Telemetry dials — Aegis coverage, Asfalya mAP and capture rate, Warden tests, and the language mix

Upstream contributions and shipped systems

portfolio · linkedin · bedometom@gmail.com


DWG SET XR-001 — XR-025 · REV 2026.10

Pinned Loading

  1. microsoft/vscode microsoft/vscode Public

    Visual Studio Code

    TypeScript 195k 44.2k

  2. nasa/openmct nasa/openmct Public

    A web based mission control framework.

    JavaScript 13.2k 1.5k

  3. aegis aegis Public

    Aegis — network-verified trust for AI agents that spend money. A CAMARA trust gate (GSMA Open Gateway, Nokia Network-as-Code) in front of every money-moving action on Google Ads and Meta.

    TypeScript