MLOps Engineer

Slingshot AI is the team behind Ash, the first AI designed for mental health. Our mission is to make support more accessible and help people change their lives in the ways they want. We’re building a world-class team by empowering individuals with the autonomy, flexibility, and support they need to do their best work. We dream big, iterate fast, and care deeply. If that sounds like you, we’d love to hear from you.

Our team spans machine learning, product, engineering, conversational design, clinical, growth, and operations, with offices in both New York City and London. We are a well-funded Series A company, having raised $93M from top-tier tech investors.

As MLOps Engineer, you’ll join our tight-knit machine learning team working on psychology foundation models. Our models have real-world impact, so this is a pragmatic, high-impact role. You’ll be able to work at a faster pace than almost anywhere else while writing high-quality code and producing meaningful scientific insights. We have a rich and growing dataset, and constantly run experiments to find the best way to use it to improve our models. Some of our current work includes:

• Data collection and curation
• Continued pre-training
• Ablation studies
• Creating synthetic datasets
• Supervising the creation of hand-crafted data
• Preference optimisation
• Training reward models
• State-of-the-art reinforcement learning research

You’ll be responsible for ensuring that our data pipelines, model training setup, and model serving infrastructure work together smoothly. You’ll also contribute to our end-user product, improving user experience through your work on our models and model orchestration. You’ll be working with the latest open-source language models as well as frontier models through our deep partnerships with the largest AI labs. You’ll read papers and identify state-of-the-art techniques for us to learn from and contribute to our core ML research. We write high-quality, typed, Zen code, mostly in Python. Our application backend is written in Kotlin and our ML stack utilizes modern tooling in the ML space, including some that we’ve developed in-house (React/Typescript).

About you:

• Experience applying deep learning frameworks (PyTorch/TensorFlow/JAX) in production, including model deployment, monitoring, and lifecycle management.
• Experience training and adapting open-source language models, with a strong focus on dataset pipelines, reproducible environments, and scalable training workflows.
• Solid software engineering fundamentals, ideally with experience in at least one non-Python language and an understanding of modern service architectures and distributed systems.
• Able to clearly explain complex ML and MLOps concepts to non-technical stakeholders.
• Enjoy a fast-paced environment and make pragmatic decisions.
• Ultimately, you’d rather prove out an idea through quick MVP code, than present a slide deck to explain it.
• Understand and appreciate that deep learning is magic!

Key responsibilities:

• Build and maintain scalable training and evaluation pipelines, ensuring data quality, reproducibility, and smooth operation across GPU clusters.
• Design, implement, and run eval systems to measure model performance, detect regressions, and automate benchmarking before models reach production.
• Develop and operate the infrastructure powering model training and inference, improving reliability, throughput, and cost efficiency.
• Stay current with SOTA ML research and identify techniques that can be integrated into robust production workflows.
• Contribute across the stack when necessary, helping integrate new models, tooling, and ML capabilities into the product, from prototype to production deployment.

What we offer:

• A chance to join a passionate tight-knit team working on something to change the world.
• Competitive compensation (we target 90th percentile).
• Travel between our NYC / London offices.
• Usual startup perks like free lunch in our offices + generous learning budget.
• Generous budget to cover your personal therapy.