Pickle Robot Company izstrādā robotikas risinājumus piegādes ķēdes, loģistikas un ražošanas nozarēm. Uzņēmuma sistēmās apvienots fiziskais mākslīgais intelekts, ģeneratīvais mākslīgais intelekts, datorredze un robotu manipulācijas tehnoloģijas, lai automatizētu fiziski smagu darbu noliktavās. Uzņēmuma vadošais autonomais robots izkrauj no pārvadājumu konteineriem preces, kas nav novietotas uz paletēm. Tas palīdz samazināt manuālā darba radīto slodzi un uzlabot noliktavu produktivitāti. Uzņēmumā strādā 11–50 darbinieku, un nesen tas noslēdza 50 miljonu ASV dolāru B sērijas finansējuma kārtu. Pickle Robot Company paplašina savas produktu iespējas un uzņēmējdarbības darbību.
Senior Machine Learning Engineer, Perception
Build vision and grasping models for Pickle Robot’s warehouse automation systems. Lead dataset development, model training, experimentation, and evaluation to improve safety and logistics efficiency.
Apraksts
- Translate picking performance goals into machine learning problems with clear predictions, labels, rewards, loss functions, and success criteria.
- Lead dataset strategy, including synthetic scene generation, distributed cloud data jobs, scene diversity filtering, and self-supervised labeling.
- Find and resolve data bottlenecks.
- Fine-tune large vision backbones with task-specific model heads.
- Plan and run controlled experiments, choosing what to test next and when to conclude.
- Create offline metrics that forecast robot performance and online metrics to improve edge-case handling.
- Review code thoroughly.
- Mentor less experienced engineers.
- Help shape the perception and grasping roadmap.
- Deliver deployable models within latency budgets, with exportable graphs and reproducible artifacts.
- Work with platform and robotics engineers to integrate runtime systems and optimize inference.
Prasības
- At least five years of machine learning and computer vision experience, including personally taking models from problem definition through measured results.
- Demonstrated ability to frame problems and turn broad product requirements into specific engineering work.
- Experience building datasets at scale through generation or collection, curation, automated quality checks, coverage and bias analysis, and active learning.
- Strong evaluation practices, including baselines, ablation studies, and metric design.
- Expert proficiency in Python and PyTorch.
- Working knowledge of cloud training infrastructure such as GCP or AWS, Docker, and experiment tracking tools.
- Professional experience in at least one area: developing and training multi-task, multimodal, or multi-view models; or self-supervised, outcome-based, or reward-driven learning.
Priekšrocības
- Health insurance
- Dental coverage
- Vision coverage
- Unlimited vacation
- Paid federal and state holidays
- 401(k) contributions equal to 5% of salary
- Company-paid travel expenses
- Additional perks intended to make work more enjoyable, comfortable, and productive