16 days ago
Dublin, IrelandStaff+
Responsibilities
- Build and train Edge AI models, including Transformers, LLMs, CNNs, and LSTMs, to process application logs, kernel traces, and multimodal data.
- Integrate machine learning flows and cloud-based LLM APIs such as Gemini, OpenAI, and Claude, with an emphasis on synthetic data creation.
- Develop algorithms to cluster log patterns and detect software regressions, race conditions, and crash precursors.
- Design supervised and unsupervised models such as Autoencoders and Isolation Forests for time-series data from CAN bus systems and onboard sensors.
- Correlate signal anomalies and system events across modalities to identify root causes.
- Port and optimize PyTorch and TensorFlow models for CPU, GPU, and embedded NPU targets.
- Apply quantization, pruning, distillation, and memory optimization for strict RAM and flash constraints.
- Define on-device data filtering and determine which data is processed locally versus in the cloud.
- Lead edge ML pipeline architecture and mentor junior engineers on embedded AI best practices.
Requirements
- Bachelor’s degree in Computer Science, Electrical Engineering, Software Engineering, or a related field.
- At least 7 years of machine learning engineering experience, including at least 3 years focused on Edge AI or embedded systems.
- Proven experience mentoring junior software engineers.
- Expert Python skills and working knowledge of modern C++, including C++14/17 for inference.
- Deep proficiency with PyTorch or TensorFlow and experience with inference engines such as ONNX, TFLite, or TVM.
- Experience with NLP techniques, sequence modeling using RNNs or GRUs, or lightweight LLMs and SLMs.
- Experience with scikit-learn, tslearn, or statsmodels for anomaly detection on sensor data.
- Ability to lead technical projects from concept through production and communicate trade-offs with stakeholders.
- Experience deploying to edge environments such as ARM-based systems, manually managing memory, and working with limited compute resources.
- Candidates with a strong computer vision track record are encouraged to apply.
- A master’s or PhD in Computer Science, Engineering, or a related field is desirable.
- Familiarity with edge systems and automotive technologies such as CAN DBC files, UDS, SOME/IP, or MQTT is desirable.
- Understanding of Linux and QNX kernel logs, process states, and operating-system-level debugging is desirable.
- Experience with NVIDIA TensorRT and Qualcomm SNPE is desirable.
Benefits
- Hybrid role based in Dublin, Ireland, with three required office days per week.
- Remote work is an option for candidates located in different geographies.
