Job Details

IT company Selecty is looking for a LEAD AI/ML ENGINEER for an International Airport Sheremetyevo project. The role involves creating an AI assistant for tour sales using LLM, RAG, and computer vision. Responsibilities include AI system architecture, recommendation quality, rapid iteration, and scaling. Requires experience with production LLM applications, prompt engineering, vector search, and Python.

We are creating an AI assistant for tour sales – a product that will change how millions of Russians choose and buy vacations. This is a startup within a large company. The product uses the latest achievements in LLM, RAG systems, and computer vision to turn the complex process of choosing a tour into a natural dialogue. The user simply describes what they want – the AI finds ideal options, considering not only text but also visual characteristics of hotels, reviews, and preferences. We are starting with the premium segment (Maldives), quickly validating the approach, and scaling to mass destinations (Turkey, Egypt). The goal by the end of the year is to prove that AI can sell tours better than traditional agents. Your role will be to be responsible for the entire AI part of the product: AI System Architecture: from selecting LLM models to building the RAG pipeline and integrating computer vision – all key technical decisions are on your side. Recommendation Quality: we don't do proof-of-concept – each iteration must increase conversion. You are responsible for ensuring that AI truly helps people find the perfect vacation. Rapid Iterations: we release changes every week, collect metrics, analyze, and improve. An ideal environment for those who want to see the results of their work in real-time. Experiments: A/B tests of prompts, hybrid search (text + visual + reviews), multimodal models – you decide what to test and quickly verify hypotheses with real users. Scaling: starting with 2000 tours to the Maldives, by the end of the year processing 50,000+ tours across three destinations with thousands of requests per day. This is not a classic ML engineering role in a large company. It's an opportunity to build a product from scratch, experiment quickly, and influence the entire architecture.

Technical Skills: Experience with production LLM applications: you have already built RAG systems or chatbots based on GPT/Claude, and know all the pitfalls. Expert-level prompt engineering: you know how to get the most out of models through the correct prompt structure, few-shot examples, chain-of-thought. Experience with embeddings and vector search: you understand how similarity search, hybrid search work, and when reranking is needed. Production-level Python: you write clean, fast code that is easy to maintain. Will be a strong plus: Experience with computer vision / multimodal models (CLIP, GPT-4o, Claude 4.5 Sonnet). Experience with high-load systems and latency optimization. Understanding of recommendation systems. Understanding of MLOps: model monitoring, versioning, A/B testing. Experience building recommendation systems. Experience building MCP for LLM systems. You have used LLM yourself to automate your tasks and know what works and what doesn't. Soft Skills: Product thinking: you understand that AI is a tool for solving business problems, not an end in itself. Ability to work with metrics: conversion, session time, escalation rate – these are your main KPIs. Speed of decision-making: it's better to release an imperfect solution in a week and test it than to spend a month making a perfect one. Ability to explain complex things in simple terms: you need to convey to the team and the business why we choose a particular architecture.

Official employment in accordance with the Labor Code of the Russian Federation. 5/2 work schedule, office (shortened working day on Fridays until 15:30). Corporate training and development programs for employees. Medical care: Voluntary medical insurance (DMS) and own medical unit. Special offers for employees from partner companies (restaurants, shops, recreation, events). Organization of employee recreation (and their children), kindergarten payment. Accredited IT company. Office at Kurskaya metro station. Corporate terms for tour purchases. Additional paid leave - 3 days, paid sick leave. DMS for family members under corporate terms. Corporate terms for learning English and sports.