AI engineer for brand content

INDG | Grip; United States; Remote; USD 140,000 - 200,000; Full time

Job summary

Grip — инфраструктура для корпоративного творческого производства, помогающая глобальным брендам генерировать контент с контролем и надёжностью. В роли Senior AI Engineer вы строите интеллектуальные цепочки: LLM‑агентов, основанных на бренд-knowledge, и конвейеры, превращающие клиентские материалы в структурированные данные, которые можно обрабатывать на уровне систем.

Ключевые задачи включают разработку надёжных агентов, настройку RAG‑проектов, управление жизненным циклом моделей и обеспечение

Qualifications

• Опыт внедрения LLM‑систем в промышленную эксплуатацию.
• ,
• Опыт разработки и валидации экспериментов в рамках качества

Responsibilities

• Проектирование и внедрение агентов, отвечающих за точное направление креативной задачи
• Настройка и верификация пайплайнов для сборки структурированных данных

Skills

LLM‑системы в продакшене

Оценочные рамки и структурированная вa

TypeScript

Python

Обучение/финетюнинг моделей

Генеративные пайплайны изображений

Архитектура и оркестрация систем

Извлечение и структура данных

Коммуникация с командами разработки ик

Tools

Elixir

C#

Job description

Описание:

Grip is changing how global brands produce content - from manual, tool-driven workflows to intelligent systems that generate high-quality visual output at scale. As a Senior AI Engineer, you build the intelligence behind that shift: LLM agents that understand what a brand looks like, retrieval systems that ground every generation in real brand knowledge, and automated pipelines that turn messy client material into structure machines can work with.

Задачи:

• designing focused, reliable agents that translate brand intent into precise creative direction crafting, testing, and versioning the prompts that steer agents and generation, and building the system that assembles them - prompting as an engineering discipline, not trial and error grounding generation in each brand's rules, assets, and visual language, with strict isolation between clients systematic measurement (RAGAS or similar), A/B testing, and regression gates, so quality is proven rather than assumed pipelines that pull structure out of documents, decks, scans, and imagery; OCR and vision models for reading, understanding, and tagging whatever a client hands us end-to-end pipelines that run without babysitting: classify, extract, propose, verify - humans review outcomes instead of doing the work by hand fine-tuning LLMs and neural networks where prompting hits its ceiling, running models locally or hosted depending on what the job needs your pipelines run against real client data and real production volume, and the results are visible in everything that ships

Требования:

• Strong experience with LLM-based systems in production (agents, RAG, orchestration)
• Hands-on work with evaluation frameworks and structured experimentation — quality you can prove, not claim
• Solid programming skills across TypeScript and Python, with experience shipping working systems
• Experience training or fine-tuning models and managing their lifecycle
• Exposure to generative image pipelines and what it takes to make their output usable (not theoretical interest — actual contact)
• In systems, not prompts — how components interact, fail, and scale
• In constraints — how to guide models toward predictable outputs
• In data — extraction, classification, and structure as the foundation everything else stands on
• Pragmatically — you ship working solutions and refine based on results
• Translating abstract creative direction into structured AI workflows
• Turning messy, unstructured material into clean, reliable data pipelines
• Debugging LLM and retrieval behavior across multiple layers of a pipeline
• Communicating across engineering and creative teams

Условия:

Grip is infrastructure for enterprise creative production—powering global brands to generate content at scale with control and consistency. AI-generated content is moving from experimentation to production. The challenge is no longer generation—it’s control, reliability, and scale. You’ll work with ComfyUI-based workflows, multi-tenant architecture, and a stack spanning TypeScript, Python, Elixir, and C#—built for production, not demos. You will define how AI is actually used in real-world creative systems—turning unpredictable models into dependable pipelines.