Software Engineer, Model Integrations

🕒 Julho 27

🇺🇸 Estados Unidos – Remoto (EUA)

⏰ Tempo Integral

🟡 Pleno

🟠 Sênior

🧑‍💻 Engenheiro Full-stack

👻 Score fantasma 26%

O que significa: quão provável esta vaga está aberta sem contratação ativa.Como ler: 60% ou mais significa que achamos que é uma vaga fantasma. Pontuações mais baixas preocupam menos.Como pontuamos: tempo em aberto, republicações, vagas abertas vs tamanho do time, rapidez para fechar vagas, e se o salário está divulgado.Observação: é uma estimativa de dados públicos, não um julgamento da empresa. Vagas reais podem ter pontuação alta, então use como uma dica, não como um fato.

Candidatar-se

Encontrar Vagas Remotas Similares

Configurar Alerta de Email para Novas Vagas

📊 Verifique sua pontuação de currículo para esta vaga

Melhore suas chances de conseguir uma entrevista verificando sua pontuação de currículo antes de se candidatar.

Enviar Currículo

Foundry

SiteLinkedInTodas as Vagas

201 - 500 funcionários

Fundada em 1996

📱 Mídia

🎮 Jogos

💰 Private equity em 2011-04

Media • Gaming

A Foundry é uma empresa de software que fornece ferramentas criativas para cinema, televisão, animação e efeitos visuais. Seu portfólio de produtos inclui Nuke (composição e revisão baseada em nós), Nuke Stage (produção virtual), Katana (desenvolvimento de aparência e iluminação), Mari (pintura e texturização 3D em alta resolução), Flix (plataforma de desenvolvimento de histórias), SmartRoto (plugin de rotoscopia com tecnologia de IA), Griptape (criador de nós visuais com IA) entre outros. A Foundry tem como alvo estúdios de VFX, estúdios de animação e equipes criativas, destacando-se pela velocidade, qualidade, fluxos de trabalho profissionais, produção virtual e ferramentas assistidas por IA. A empresa oferece licença educacional e seu software é amplamente utilizado pelos principais estúdios, tendo sido presente em trabalhos vencedores do Oscar de VFX.

Descrição

• Ship model support on day one. When a major new foundation model, video generator, voice model, or open-weight checkpoint drops, artists should be able to try it in Griptape Nodes immediately. You build the node and library that makes that real. • Translate model behavior into artist language. When a model fails, the system knows what went wrong; the artist usually doesn't. Your job is closing that gap — turning "403: proxy model failed with [800 lines of garbage]" into "Your reference image needs to be at least 1024px wide and 32-bit; yours is 859px and 24-bit." Client-side validation, crisp error messages, sensible defaults, predictable behavior — the craft that turns "technically functional" into "actually pleasant." • Design the next-gen interface for open-source and local models. Architect how Griptape Nodes treats self-hosted and on-device models — and build the deep local-model support that lets customers run chat, agent loops, and more entirely locally, including the substrate Griptape Studio and other parts of the platform will lean on. • Stay responsive to two signal streams. New models drop nearly every week. Artist feedback never stops. Build the muscle and the systems to turn both into shipped improvements quickly. • Make AI part of how we build. Use Claude, Cursor, Griptape Nodes itself, and whatever else clears the path. We expect engineers here to ship faster and reason deeper because of AI, not in spite of it. • Hold the quality bar. Design, implement, test, and document features for the way studios actually work. Review code thoughtfully, estimate honestly, and partner with Customer Support when issues surface from the field.

🎯 Requisitos

• Strong Python with the engineering habits to back it up — type hints, tests, packaging, and dependency management aren't optional in your code. You can own a non-trivial feature end-to-end. • A degree in computer science. • Customer empathy and intuition. You can write the validation, error messages, and defaults that translate technical reality into language an artist can act on — without needing to ask an artist what to write. • Working knowledge of the modern model ecosystem — foundation models, diffusion, video and audio generators, model hubs (Hugging Face and friends), and what's involved in running models locally vs. through a hosted API. • Hands-on experience with local inference tooling — Ollama, llama.cpp, vLLM, MLX, transformers, or equivalent. You've actually gotten models running on someone's laptop. • Generative AI fundamentals — diffusion models, LLMs and agents, fine-tuning, MCP, evaluation metrics, prompt engineering. Enough to debug a workflow when it goes sideways. • A track record of using AI tools (Claude, Cursor, Copilot, agent frameworks) to genuinely accelerate your work, not as a novelty. • Strong written communication — clear design docs, useful PR descriptions, code comments that future-you will thank past-you for. • Comfort working across Windows, Linux, and macOS. • ****Nice to have, keen to learn:**** • Experience shipping AI/ML integrations inside a product environment. • Familiarity with media and entertainment workflows and formats — OpenEXR, USD, ACES, OpenColorIO, common containers and codecs. • Open-source contributions to ML or AI tooling. • Working knowledge of model quantization, LoRA fine-tuning, or production model evaluation. • Background in node-based, dataflow, or visual programming environments. • Experience integrating with DCC tools (Nuke, Maya, Houdini, Blender).

🏖️ Benefícios

• Health Savings Account & 401K Plan • Medical, Dental and Vision plans • Bonus scheme • Anniversary day off • Passion Days • Annual personal learning & development time