Skip to content
View Goktug-Dusunen's full-sized avatar
:electron:
:electron:

Block or report Goktug-Dusunen

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
Goktug-Dusunen/README.md

Göktuğ Düşünen

AI Engineer · Turkish NLP · Retrieval · Speech · Vision · Security

Live Portfolio · Hugging Face · Werea · LinkedIn · Email

I build open AI systems for Turkish language understanding and production use. My work spans compact retrieval models, speech and text restoration, OCR, edge vision, and defensive security. I publish models, datasets, benchmarks, and interactive demos through my personal Hugging Face profile and Werea.

Selected open-source work

Project Focus Explore
DÜŞÜNEN Retrieval Stack Turkish embeddings, Matryoshka representation learning, reranking, hard negatives, and evaluation Rota 270M · Atlas 278M · Benchmark
Werea Turkish NLP Diacritics restoration, summarization, QA, NER, intent, topic, sentiment, and NLI TextRestore · Summarizer · QA
Turkish Speech & OCR ASR, TTS, printed-text OCR, and document understanding Dinle ASR · Oku OCR · Werea TSS · DocOCR
Security AI SOC assistance, incident analysis, defensive workflows, and security operations interfaces NanoSOC · Argus · Security Console
Edge Vision Person detection for edge accelerators and constrained inference environments Gemstone Detector

Live demos

Engineering focus

  • Modeling: PyTorch, Transformers, Sentence Transformers, PEFT/QLoRA
  • Retrieval: dense embeddings, reranking, hard-negative mining, RAG evaluation
  • Deployment: ONNX, quantization, edge inference, Gradio, Hugging Face Spaces
  • Domains: Turkish NLP, speech, document AI, computer vision, blue-team security

Beyond models

I also work on device-level product experiments. My customized Glyph Developer Kit explores real-time delivery progress on Nothing Phone's Glyph Interface.

Open portfolio

The public portfolio currently includes 30 models, 20 datasets, and 9 Spaces under GoktugD, with additional organization releases under Werea.

The machine-readable catalog and weekly synchronization workflow live in open-ai-portfolio.

Open to research collaboration, applied AI work, and technically ambitious product teams.

Pinned Loading

  1. Glyph-Developer-Kit Glyph-Developer-Kit Public

    Forked from Nothing-Developer-Programme/Glyph-Developer-Kit

    Real-time Turkish delivery tracking on Nothing Phone Glyph Interface with tiered LED progress.

    Kotlin 8