AI Engineer (Middle/Senior)

We're looking for an AI Engineer, middle or senior, to build the AI in our products: assistants and agents, retrieval over documents, and speech and language models for Uzbek and Russian.

What you will do

• Build LLM applications end to end: RAG, tool-using agents, structured extraction and their evaluation
• Fine-tune and evaluate models (LoRA/QLoRA), including for Uzbek and Russian
• Serve models efficiently (vLLM or similar, quantization, batching) and keep latency and cost in check
• Build evaluation sets and metrics, run experiments and turn the results into product decisions
• Ship AI features behind APIs together with backend engineers, and monitor them in production
• Senior: lead the technical decisions of a product area and mentor other engineers

Requirements

• 2+ years building ML/AI systems that reached production (4+ for senior)
• Strong Python; PyTorch (or JAX/TensorFlow)
• Hands-on LLM work: prompting, RAG, embeddings and vector search, fine-tuning or evaluation
• Solid ML fundamentals: training, validation, metrics, overfitting, data leakage
• Deployed models behind an API (FastAPI, vLLM, Triton or similar); Docker, Linux, Git
• Uzbek, Russian or English for daily work; you read research papers in English

Nice to have

• Speech (ASR/TTS) or NLP for low-resource languages
• Agent frameworks (LangGraph or similar), GPU optimisation, Kubernetes
• Publications, open-source models or strong Kaggle results

Hiring process

1. CV review
2. Online test (40 minutes)
3. Call with HR (20 minutes)
4. Take-home: a RAG assistant with a measured evaluation (8–10 hours, 6 days to submit)
5. Technical deep-dive (90 minutes, online)
6. ML system design (60 minutes, online); this is where we settle middle or senior
7. Meet the team (45 minutes, at our Tashkent office)
8. Offer

2+ years shipping ML/AI to production (4+ for senior); Python, PyTorch; LLMs: RAG, fine-tuning, evaluation; model serving.