机器学习工程师
ML Engineer
Orcrist 正在使用前沿技术构建下一代数据智能平台。我们处理海量数据，实现亚秒级查询。我们的产品是一个基于 Kubernetes 的平台，以 B2B SaaS 或自托管本地部署方案提供，包括隔离网络部署。我们为国防、执法和企业客户提供服务，帮助他们将关键任务数据转化为可操作的智能。
职位描述
我们正在寻找一位实战型机器学习工程师，负责在文本、视觉、音频和其他应用机器学习用例中构建和生产化现代 AI 能力。你将直接与最前沿的开源模型合作——进行测试、评估、优化和微调，以满足实际产品用例的需求。这不是一个专注于从零训练基础模型或仅构建数据管道的角色。你将与研究、产品和数据工程团队紧密合作，将有前景的模型和想法从实验阶段转化为可靠、可生产系统的功能。
你将负责的工作
- 评估、比较并集成开源模型，以满足具体的产品和客户用例。
- 构建和改进基于大语言模型的应用。
- 在适当的情况下，通过推理优化、评估和微调来扩展和改进 AI 和 ML 模型。
- 与自然语言处理、翻译、语音转文字 / ASR、图像和文档理解以及其他相关应用 AI 模型协作。
- 设计涵盖模型质量、延迟、可靠性和成本的评估框架。
- 将模型从实验阶段带入生产环境，包括打包、部署、监控和迭代。
- 优化推理性能和运营成本。
- 与研究和产品团队合作，将原型和实验转化为可扩展的产品能力。
- 在保持对模型和模型行为动手实践的同时，为现代机器学习基础设施和 MLOps 做出贡献。
我们希望你具备
- 4 年以上机器学习工程、应用 AI 或类似实战型 ML 相关经验。
- 强大的 Python 技能，以及使用现代 ML 框架和库（如 PyTorch、Transformers 和 Hugging Face）的实际经验。
- 具备使用大语言模型、自然语言处理模型、语音模型或其他现代生成式 AI 系统的经验。
- 有实际操作现有模型的评估和实验经验，而不仅仅是构建 ML 基础设施。
- 有使用推理引擎（如 vLLM 或 SGLang）和平台（如 NVIDIA Triton 或 Ollama）的经验。
- 熟悉微调、提示工程、模型压缩等技术。
查看英文原文
Orcrist is building a next generation data intelligence platform using cutting-edge technologies. We're handling petabyte-scale data with sub-second queries. Our product is a Kubernetes‑based platform delivered as B2B SaaS or as a self‑hosted on‑prem solution, including air‑gapped deployments. We enable customers across defense, law enforcement, and enterprise to turn mission-critical data into actionable intelligence.
Role
We are looking for a hands-on ML Engineer to build and productionize modern AI capabilities across text, vision, audio, and other applied ML use cases. You will work directly with state-of-the-art and open-source models — testing, evaluating, optimizing, and fine-tuning them for real product use cases. This is not a role focused on training foundation models from scratch or building data pipelines only. You will work closely with Research, Product, and Data Engineering teams to take promising models and ideas from experimentation to reliable, production-ready systems.
What you’ll do
- Evaluate, compare and integrate open-source models for concrete product and customer use cases.
- Build and improve LLM-based applications.
- Extend and improve AI and ML models using inference optimization, evaluation, and fine-tuning where appropriate.
- Work with NLP, translation, speech-to-text / ASR, image and document understanding, and related applied AI models.
- Design evaluation frameworks covering model quality, latency, reliability, and cost.
- Take models from experimentation into production, including packaging, deployment, monitoring, and iteration.
- Optimize inference performance and operational cost.
- Collaborate with Research and Product teams to turn prototypes and experiments into scalable product capabilities.
- Contribute to modern ML infrastructure and MLOps while remaining hands-on with models and model behaviour.
About you
- 4+ years of experience in Machine Learning Engineering, Applied AI, or a similar hands-on ML role.
- Strong Python skills and practical experience with modern ML frameworks and libraries such as PyTorch, Transformers, and Hugging Face.
- Experience working with LLMs, NLP models, speech models, or other modern generative AI systems.
- Hands-on experience evaluating and experimenting with existing models rather than only building ML infrastructure.
- Experience with inference engines like vLLM or SGLang, and platforms like NVIDIA Triton or Ollama.
- Familiarity with fine-tuning, prompting, model evaluation, inference, and deployment.
- Working understanding of AI model encoding and quantization formats, and the tradeoffs thereof.
- Strong engineering mindset and ability to turn experiments into reliable, reproducible systems.
- Eligible to work in Germany; export-control screening required for certain programs.
Nice-to-haves
- Hands-on experience with model serving and production deployment, using technologies such as Kubernetes, KServe, NVIDIA Triton, and/or Ray Serve.
- Knowledge of inference optimization techniques, including batching, quantization, ONNX, or TensorRT.
- German language skills (B1+) and/or familiarity with defense or public safety datasets.
- Exposure to geospatial AI, satellite imagery, or remote sensing.
- Experience working in constrained or regulated environments with infrastructure, security, or deployment requirements.
What we offer
- Remote-first, Germany-wide: Work from wherever you do your best work, with regular team gatherings in Berlin and other off-site locations.
- Flexibility by default: Flexible working hours help you make work fit your life.
- Your setup, your way: Get a personal home-office equipment budget to create a workspace that works for you.
- 30 days of vacation: Take the time you need to recharge and come back with fresh energy.
- Keep growing: We invest in your personal and professional development.
- Get rewarded for impact: Performance bonuses are tied to agreed objectives and key results.
- A warm welcome: Every new team member gets a welcome goodie bag.
- Good people, good times: From summer and Christmas parties to regular team gatherings, we make time to celebrate together.
- A mission that matters: Work on challenges with tangible impact on public safety and national security.