Build perception algorithms and architecture, fusing multiple sensor modalities and supporting an evolving set of novel object classes, in order to provide safe operations for heavy machinery working on construction and mining sites.
Core Responsibilities- Re-use, adapt and extend existing perception algorithms (such as those commonly used in AV applications) to be applied to Teleo's operational domain and develop novel multi-modal perception algorithms
- (Re-)Train models as Teleo's operational domain evolves
- Develop auto-annotation and auto-labeling tools using SoTA methods, such as VLMs
- Define evaluation protocols that correlate with on-ground performance
- Drive active learning: select the right data
- Integrate tightly with MLOps for continuous deployment
Required Qualifications- M.S. or higher in Computer Science, Computer Engineering, Robotics, Electrical Engineering, or a related technical field.
- 3+ years in applied ML with large-scale datasets
- Strong Python + PyTorch
- Experience shipping perception or ML systems into production
- Systems thinker: understands data, models, infra as one system
Strong Experience With- Auto-annotation techniques like VLM-based labeling
- Model evaluation beyond single metrics (failure modes, edge cases)
- Perception tasks: detection, segmentation, depth, tracking
- Multi-modal data (camera, LiDAR, radar)
Nice to Have- Dataset management & slicing
- Experience with synthetic data or simulation
- Large-scale data pipelines (Parquet, Arrow, object storage)