Omnisent is pioneering scalable acoustic sensing powered by AI — transforming audio into real-time intelligence.

We build hardware + software tailored for acoustic data, designed to extract insights from complex acoustic signals. Our proprietary ultra-low-power sonic devices capture and process acoustic signals in real-time, training what we call a Large Acoustic Model (LAM) — our foundation model built to decode the rich, messy, and untapped world of non-speech audio.

We’re starting with the manufacturing industry, applying our tech to compressed air systems — one of the most overlooked sources of energy waste. From there, we’re expanding into energy, defense, space, and smart cities — sectors where sound is the next frontier.

Please note: This is an on-site position in Munich. Applicants must have a valid EU work permit and be available to work from our Munich office during the semester.

Aufgaben

Tune hyperparameters and optimize classification pipelines for classical Machine Learning (ML) models and Deep Neural Networks (DNNs)
Extract and engineer features from raw acoustic signals (e.g. FFTs, spectral and time-domain features)
Compare models using suitable quantitative metrics and interpretability methods (i.e. XAI tools)
Design and test neural network architectures
Analyze model behavior on unseen or noisy data, and test generalization
Maintain code quality through modular design, logging, and documentation
Prototype lightweight model deployment (e.g., exporting models, running inference APIs)
Create visualizations to understand feature behavior, model decisions, and inference quality

Qualifikation

Basic Requirements:

Currently enrolled in a Master’s program in Computer Science, Machine Learning, Data Science, or related field
Strong Python skills, including experience with: numpy, scikit-learn, matplotlib, pytorch
Experience training and testing ML and DL models
Experience using Git, writing clean, modular, and reproducible code
Solid understanding of model evaluation and basic interpretability (confusion matrices, feature importance, Grad-CAM, etc.)
Curious, self-driven, and structured in your approach
Fluent in English

Preferred Skills:

Experience working with acoustic data, time-series, or Fourier/spectral representations
Familiarity with feature extraction for audio: time-domain (e.g., peak analysis, RMS) and frequency-domain (e.g., FFTs, spectral features)
Experience with model interpretability (such as GradCam, Shap, LIME, feature importance, etc.)
Experience with model deployment for real-time applications and/or cloud services such as Azure

Join us in redefining how the world listens, understands, and responds through intelligent acoustic sensing.

In your cover letter, we’d love to hear about your personal interests, what drives you, why this role feels like the right fit — and most importantly, why Omnisent.

Helpful information

Job recommendations