Skip to content
View Gjusev's full-sized avatar
🏠
Working from home
🏠
Working from home

Highlights

  • Pro

Block or report Gjusev

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. Markdown is supported. 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
Gjusev/README.md
Animated Medusa turns from profile to meet the viewer's gaze

Voice agents, open-source AI tools & independent software.

I'm Youssef, founder of mokka, based in Siegen, Germany.
I build tools to test AI systems, make their decisions inspectable, and keep data under your control.

Selected work · AI tooling · More projects · Get in touch

Siegen, DE / web · AI · privacy / palestina libre


Selected work

Open-source projects with code, examples, and a way to try them.

voice-evals: replay scoring, live probes and inspectable session artifacts

Replay calls or probe a live WebSocket agent. Gate on transcription, latency, interruptions, and task outcomes.

Python · CLI + library · Apache-2.0

Code · PyPI · Demo · Kaggle

note-lm: a local research notebook with sources and traceable evidence

A local research notebook with versioned sources, cited claims, and evidence you can trace back to the original passage.

Windows desktop · Local storage and models · MIT

Code · Download · Demo

Local support-ticket routing with urgency signals and human handoff. Includes a fine-tuned BANKING77 model and reproducible evaluations.

Python · Local decision model · Apache-2.0

Code · Try the app · Model · Results

A German-speaking voice agent that qualifies heating-service leads, with a dashboard and a two-agent call simulator.

Next.js · ElevenLabs · PostgreSQL · MIT

Code · Live demo

AI tooling

Small Python packages for routing, retrieval, and evaluation. The laya tools use a local decision model; the Clef tools integrate with Cloudflare's decision models.

Task Local · laya Cloudflare · Clef
Evaluate — confidence audits & CI gates laya-evals · PyPI clef-evals · PyPI
Route — choose an LLM per request laya-router · PyPI clef-router · PyPI
Compact — fit evidence to a token budget laya-compactor · PyPI clef-compactor · PyPI

laya-phishield applies the same approach to phishing detection: local semantic signals, deterministic email checks, and an explainable risk score. PyPI

These are independent tools built on laya and Cloudflare Clef. Each repository documents its setup, benchmarks, and current limitations.

Document & vision evaluation

Your extractor reads the same invoice in three locales. Your vision API gets the same pixels. Measure both — offline-first regression kits with deterministic CI gates.

More projects

Application code, architecture notes, and demos. Status reflects each repository's current scope.

Browse all repositories →

Working with

Build — Python, TypeScript, React, Next.js, FastAPI, Tauri.
Data & infrastructure — PostgreSQL, Qdrant, Docker, n8n.
AI systems — voice agents, RAG, local models, evaluation, Playwright testing.


mokka — websites, AI products, and automation for small businesses.
Also building klarbescheid for understandable benefits notices and datenmaske for PDF redaction.

Let's build something useful →

Pinned Loading

  1. laya-router laya-router Public

    OpenAI-compatible proxy that routes prompts to cheap or frontier LLMs using a local System 1 decision model (laya)

    Python 2

  2. laya-triage laya-triage Public

    Multilingual support ticket triage with a local System 1 decision model: hierarchical intent routing, urgency/frustration/churn signals, confidence-based escalation

    Python 1

  3. clef-compactor clef-compactor Public

    Query-aware RAG context compaction with Cloudflare Clef: keep the evidence, cut noisy retrieval context.

    Python 1

  4. clef-router clef-router Public

    Route prompts between cheap and frontier LLMs with Cloudflare's Clef decision model. OpenAI-compatible proxy + library.

    Python 2 1

  5. nexary-platform nexary-platform Public

    On-premise enterprise AI platform: multi-provider LLM chat, hybrid RAG (dense + BM25 + cross-encoder rerank), SYSLOG-auditable — runs inside customer networks

    TypeScript 1

  6. voice-evals voice-evals Public

    Reproducible evaluation for voice agents: replay scoring, scripted live probes over WebSocket, and CI gates for WER, latency, barge-in behavior and task outcomes.

    Python 1