ML Internship Deep Learning for Causal Inference - ethon

Ethon was built to eliminate the trillion-dollar waste problem in manufacturing. We are the technology leader in Industrial AI, building agentic workflows that help manufacturers optimize cost, quality, and speed of production. We are a fast-growing team driving massive impact in one of the largest and most under-served industries in the world.

We serve the world’s top manufacturers across the globe. From Siemens and Bosch to Lindt and Roche, our customers trust us to make their factories run better. We are backed by leading investors like Index Ventures, General Catalyst, and Earlybird. Our team draws from Google, Meta, Palantir, MBB, and top manufacturing companies, all here for the same reason: giving manufacturers the technology to run every factory at unprecedented productivity.

What is this internship about?

We are looking for a machine learning intern with strong engineering skills and the drive to turn ambitious ideas into solid, working prototypes.

In this internship you will work at the intersection of modern deep learning and industrial applications like Root Cause Analysis (RCA). Your mission is to explore how deep learning can be leveraged to infer causal structure and drivers of variation directly from factory data. Concretely, you will:

Investigate modern deep learning approaches for causal inference on tabular data, and evaluate which directions are most promising for industrial settings.

Design approaches that jointly exploit tabular signals (sensor readings, process parameters, quality measurements) and contextual information in natural language (sensor names, asset hierarchies, process descriptions, other factory metadata) to improve causal discovery and Root Cause Analysis.

Build the experimental and benchmarking infrastructure needed to compare your prototypes against established causal discovery baselines, on both public datasets and (anonymized) real manufacturing data.

Work closely with our ML Scientists and engineers to shape the path from prototype to a component that could augment our analytics stack.