Senior Machine Learning Engineer

Kompaniyamiz AI yo‘nalishini rivojlantirish doirasida NLP, NLU, LLM, OCR, STT, TTS va embedding modellar asosida real biznes jarayonlarini avtomatlashtirishga qaratilgan yechimlarni ishlab chiqmoqda. Asosiy e’tibor tayyor AI servislaridan foydalanishga emas, balki custom ML modellarni yaratish, train/fine-tuning qilish, baholash va production muhitga chiqarishga qaratiladi.

Rol haqida

Biz Senior Machine Learning Engineer izlayapmiz. Nomzod NLP/NLU, LLM, OCR va embedding bilan chuqur ishlay olishi, dataset tayyorlashdan boshlab model training, evaluation, deployment va monitoringgacha bo‘lgan jarayonni end-to-end boshqara olishi kerak.

STT va TTS yo‘nalishlarida tajriba bo‘lsa, ustunlik hisoblanadi.

Majburiyatlar

LLM modellarni domain-specific dataset asosida fine-tuning qilish.

SFT, LoRA/QLoRA, instruction tuning va domain adaptation pipeline’larini ishlab chiqish.

NLP/NLU modellarini yaratish: intent classification, entity extraction, text classification, question-answering va semantic search.

OCR va Document Understanding yo‘nalishida custom model yaratish, train yoki fine-tuning qilish.

Scanned document, PDF, form, table va rasmli hujjatlar uchun OCR pipeline ishlab chiqish.

Text detection, text recognition, layout analysis va post-processing jarayonlarini yo‘lga qo‘yish.

Embedding modellarni tanlash, fine-tuning qilish va domain-specific retrieval sifatini oshirish.

RAG tizimlari uchun chunking, embedding, vector search, hybrid retrieval va reranking mexanizmlarini ishlab chiqish.

Dataset tayyorlash, data cleaning, augmentation va annotation quality nazoratini amalga oshirish.

Model evaluation uchun metrikalar, benchmark va test datasetlar ishlab chiqish.

Model sifatini accuracy, latency, stability va resource consumption bo‘yicha optimallashtirish.

ML modellarni API yoki microservice sifatida production muhitga chiqarish.

Training pipeline, experiment tracking, model registry va deployment jarayonlarini tashkil qilish.

Ichki tizimlar bilan ML/NLP/OCR/LLM yechimlarini integratsiya qilish.

Texnik arxitektura tanlashda ishtirok etish.

Talablar

• Python bo‘yicha kuchli bilim va production-ready kod yozish tajribasi.
• PyTorch bilan chuqur amaliy tajriba.
• Hugging Face Transformers ekotizimi bilan ishlash tajribasi.
• LLM fine-tuning bo‘yicha real amaliy tajriba.
• SFT, LoRA/QLoRA, PEFT, instruction tuning yoki domain adaptation bo‘yicha bilim.
• NLP/NLU modellarini train yoki fine-tuning qilish tajribasi.
• OCR yoki Document Understanding modellarini yaratish yoki fine-tuning qilish tajribasi.
• Embedding models, semantic search, vector search va retrieval modellarni yaxshi tushunish.
• RAG, hybrid search, reranking va vector database bilan amaliy tajriba.
• Dataset tayyorlash, preprocessing, augmentation va evaluation jarayonlarini mustaqil olib bora olish.
• Model training, fine-tuning, evaluation, deployment va monitoring jarayonlarini end-to-end yurita olish.
• Docker, Linux, Git va FastAPI bilan ishlash tajribasi.
• Model sifati, inference tezligi, resurs sarfi va scalability o‘rtasidagi trade-offlarni tushunish.

Plus bo‘ladi

• Low-resource tillar uchun model train yoki fine-tuning qilish tajribasi.
• STT va TTS yo‘nalishida tajriba:
• GPU training, distributed training yoki mixed precision training tajribasi.
• Model optimizatsiyasi: quantization, batching, caching, pruning. vLLM, TensorRT-LLM yoki boshqa model serving texnologiyalari bilan tajriba.
• CI/CD va production ML monitoring bo‘yicha tajriba.
• Katta hajmdagi hujjatlar bilan ishlaydigan OCR/RAG tizimlar yaratish tajribasi.

Biz kutayotgan nomzod

Bizga tayyor modelni shunchaki chaqirib ishlatadigan emas, balki modelni yaratish, train/fine-tuning qilish, baholash va productionga chiqarish jarayonini to‘liq yurita oladigan mutaxassis kerak.

Ideal nomzod dataset sifati model natijasiga qanday ta’sir qilishini tushunadi, LLM, OCR, NLP va embedding modellarni real biznes vazifalarga moslashtira oladi, end-to-end ML pipeline qura oladi va research yoki open-source yechimlarni production darajasiga olib chiqa oladi.

Sharoitlar

• AI yo‘nalishida real production loyihalarda ishlash imkoniyati.
• NLP, LLM, OCR, Embedding va Voice AI bo‘yicha zamonaviy texnologiyalar bilan ishlash.
• Katta hajmdagi real ma’lumotlar va murakkab biznes jarayonlari bilan ishlash imkoniyati.
• Texnik qarorlar qabul qilishda faol ishtirok etish.
• Professional jamoa va o‘sish imkoniyati.
• Ish formati: full-time.
• Ish grafigi va ish haqi (3-4k$++) suhbat asosida kelishiladi.