Antimetal is building the future of infrastructure management by creating a platform that investigates, resolves, and prevents issues, giving engineers their time back to focus on building great products. The company is looking for a Research Engineer to push the boundaries of AI systems that power its core product.
Requirements
- 3–6 years of engineering or applied research experience, ideally in ML/AI systems.
- Strong software engineering background with expertise in Python and at least one deep learning framework (PyTorch or TensorFlow).
- Experience with large-scale training or fine-tuning of ML models (LLMs or multimodal a plus).
- Familiarity with distributed systems, accelerators (GPUs/TPUs), and data pipelines.
- Exposure to interpretability, robustness, or AI safety research.
- Experience with multimodal models (text + images, logs, or other data types).
- Track record of contributions to ML research (open-source repos, papers, workshops).
Responsibilities
- Design, run, and analyze experiments that advance our AI powered engine.
- Prototype new model architectures, algorithms, and data processing pipelines, balancing cutting-edge ideas with practical engineering constraints.
- Optimize distributed training and inference infrastructure for performance, reliability, and scale.
- Build tools, visualizations, and dev workflows that accelerate research velocity across the company.
- Collaborate with product and engineering teams to translate ambiguous research ideas into production-ready systems.
- Contribute to both low-level optimizations (GPU/memory efficiency) and high-level model design.
- Share findings through clear technical writing and presentations that shape Antimetal’s technical roadmap.
Other
- Comfort with ambiguity — you can design experiments, evaluate trade-offs, and iterate quickly.
- Excellent communication skills and the ability to work across research and engineering teams.
- Identify as a builder.
- Are excited to work in-person from our new and spacious office in New York.
- Love working in a startup environment (experience in a startup or obsession with going zero-to-one).
