What are the responsibilities and job description for the Junior AI Scientist position at Lensa?
Lensa is a career site that helps job seekers find great jobs in the US. We are not a staffing firm or agency. Lensa does not hire directly for these jobs, but promotes jobs on LinkedIn on behalf of its direct clients, recruitment ad agencies, and marketing partners. Lensa partners with DirectEmployers to promote this job for Schlumberger. Clicking "Apply Now" or "Read more" on Lensa redirects you to the job board/employer site. Any information collected there is subject to their terms and privacy notice.
Build, train and deploy large-scale, self-supervised "foundation" models that learn rich representations of seismic data (ND(2D,3D,...) volumes), to be fine-tuned for tasks such as event detection, subsurface imaging, fault characterization or reservoir property estimation.
If you have questions about this posting, please contact support@lensa.com
Build, train and deploy large-scale, self-supervised "foundation" models that learn rich representations of seismic data (ND(2D,3D,...) volumes), to be fine-tuned for tasks such as event detection, subsurface imaging, fault characterization or reservoir property estimation.
- Domain Knowledge
- Seismic theory & processing: data formats (SEG-Y/D), de-noising, deconvolution, stacking, migration, tomography, inversion.
- Reservoir geomechanics, rock physics, well-log integration, AVO/AVA analysis.
- Geostatistics: variography, kriging, co-kriging, uncertainty quantification.
- Machine-Learning & Foundation-Model Expertise
- Self-supervised and semi-supervised learning: masked autoencoders (MAE), contrastive methods (SimCLR, BYOL), clustering-based (DINO), predictive coding.
- Model architectures: 1D/2D/3D CNNs, Vision/Audio Transformers, graph neural networks, diffusion/generative models, multi-modal encoders.
- Transfer learning & fine-tuning at scale: prompt/adapter-based techniques, domain adaptation.
- Evaluation metrics: geophysical error norms (L2, semblance), detection/segmentation metrics (IoU, F1), end-use KPIs (horizon-picking accuracy, attribute classification).
- Software & Infrastructure
- Programming: expert Python (NumPy, SciPy, Pandas), C /CUDA for performance kernels.
- Deep-learning frameworks: PyTorch (Lightning, Distributed), TensorFlow/Keras, JAX/Flax.
- Large-scale training: multi-GPU, multi-node, mixed-precision, ZeRO optimization.
- Data engineering: advanced segmentation of terabyte-scale seismic volumes.
- Mathematical & Algorithmic Foundations
- Linear algebra, probability & statistics, optimization (stochastic, convex, non-convex).
- Signal processing: Fourier/Wavelet transforms, filtering, spectral analysis.
- Numerical methods for PDEs, inverse problems, regularization techniques.
- Collaboration & Communication
- Cross-disciplinary teamwork with geoscientists, software engineers, product managers and end-users.
- Clear presentation of complex model behaviors, uncertainty quantification and business impact.
- Desired Extras
- Contributions to open-source seismic/ML projects or standards bodies (e.g. SEG, OSDU).
- Cloud & DevOps: AWS/GCP/Azure (S3, EC2/GPU, Batch/ML Engine), Kubernetes, Terraform, Docker, CI/CD pipelines.
- Experiment tracking & MLOps: MLflow, Weights & Biases, Neptune, Grafana, Prometheus.
- Multi-modal fusion: combining seismic with well logs, production data, satellite/inSAR.
- Agile/Scrum practices: sprint planning, peer code reviews, documentation (API specs, best practices).
- Education & Experience
- PhD (or M.S. 5 years) in Geophysics, Seismology, Computer Science, Electrical Engineering, Applied Math, or equivalent.
- 2-3 years hands-on seismic/geophysical data processing or interpretation.
- Peer-reviewed publications or patents in seismic AI, geophysical inversion or related fields a plus.
If you have questions about this posting, please contact support@lensa.com