At kausable, we build causal, reasoning-first models that learn from a handful of examples and generalize across domains. Research gets us to a capable, versatile foundation model. It's great at deducing the mechanics that govern a system; yet, standalone, it's just one query-answer cycle.

This role harnesses the full potential of that core intelligence by embedding it in a sophisticated agentic system capable of tackling multi-step, long-horizon tasks.

As our Harness Engineer, you own the path from a raw ML model to an intelligent, largely autonomous system: harnesses, inference algorithmics, context curation, and hierarchical querying. You will work in the seam between research and product, giving free reign to your creativity when designing the system and exercising discipline and diligence when handing it over to the productization team.

Tasks

Turn research models into sophisticated agentic systems capable of a large degree of autonomy.
Build harnesses, inference algorithms and systems, smart context curation.
Build release criteria that show quantitatively when an agent is ready to ship.
Build developer-facing abstractions around our agentic systems.
Work closely with researchers to expose failure modes; receive product feedback from customers and turn it into better systems.
Translate customer and design-partner needs into reusable platform capabilities rather than one-off solutions.

Requirements

A track record of ML-research or strong, research-adjacent ML engineering.
Experience with in-context learning, building agentic systems, applying ML models in complex, dynamic environments.
Strong software engineering skills in Python and hands-on fluency with PyTorch.
Sound judgment around evaluation, observability, reliability and production trade-offs.
The ability to work directly with customers, researchers and product stakeholders.
A pragmatic, outcome-oriented mindset: you optimize for dependable capabilities that users can actually adopt.
We are primarily hiring at senior level. We are also open to exceptional candidates with fewer years of experience who can demonstrate comparable depth, judgment and ownership.

Nice to have:

PFNs, synthetic data or probabilistic models.
Weights & Biases, model registries, CI for models or comparable MLOps tooling.
SDK or developer-tooling design.
Prior startup, design-partner or 0-to-1 product experience.
Own model releases, monitoring and rollback patterns as the production footprint grows.
Experience with model serving, APIs, containers and cloud infrastructure.

Benefits

🚀 Where This Can Go

You will define how kausable ships ML: the patterns, tooling and standards between research and production. As the team grows, the role can expand into technical ownership of the model-to-product stack or leadership of a small ML product group. The trade-off is part of the job: shipping quickly matters, but only when the resulting system remains measurable, reusable and dependable.

🫂 Our Culture

We are "Putting Science at the Core of AI". That means we:

are scientists at heart, with a builder's mindset,
are open to challenge, grounded in curiosity and respect,
welcome diverse perspectives and value thoughtful, open debate,
focus on outcomes and real-world impact,
foster an environment of support, inspiration, and freedom for everyone to do their best work.

🏆 Perks & Benefits

VSOP equity: a real stake in what we build.
30 days of paid holiday per year.
Statutory social insurance.
Conference travel and role-relevant learning.
Flexible hybrid work, with roughly one in-person team meet-up per month.
A high-end laptop and access to the cloud compute required for the role.

⚒️ Tools and Infrastructure

Python and PyTorch.
Weights & Biases and model-evaluation tooling.
Docker, AWS, RunPod and comparable cloud infrastructure.

🫶 Sounds like it's for you? Send us your favorite way to drink coffee along with your CV or LinkedIn, and we'll get back to you soon.

If it's a match, we'll get to know each other over a number of online interviews, followed by an onsite day where we go in depth.

We are looking forward to hearing from you!