Pickle Robot Company is building perception and grasping capabilities for real-world robotics, and this role sits at the center of the modeling loop. You will own perception-focused machine learning work end-to-end, from turning picking goals into measurable learning tasks to producing deployable models that hand off cleanly to runtime and edge deployment teams. The team structure keeps platform and robotics engineers focused on integration while you focus on problem formulation, data, training, and evaluation.
What you will do
- Own ML problem formulation: translate ambiguous picking performance goals into well-defined modeling targets. Decide what the model predicts, how labels and rewards are defined, which loss to optimize, and what “better” means in practice.
- Build and improve the data engine: lead dataset strategy from synthetic scene generation and distributed cloud dataset jobs to scene diversity filtering and self-supervised label generation.
- Fix bottlenecks in data: identify when performance limits come from data rather than the model, then drive the changes needed to improve results.
- Train and iterate on large vision models: fine-tune large backbones using multiple task-specific heads, run disciplined experiments, and determine the next experiment and the right stopping point.
- Own evaluation and verification: define offline metrics that correlate with on-robot outcomes and online metrics that help you improve performance on edge cases.
- Drive technical excellence: participate in rigorous code reviews, mentor engineers earlier in their careers, and contribute to the perception and grasping roadmap.
- Hand off cleanly to production: deliver deployable models with reasonable latency budgets, exportable graphs, and reproducible artifacts, partnering with engineers who own runtime integration and inference optimization.
What you bring
- 5+ years of experience in machine learning and computer vision, including models you carried from problem statement to measured outcome.
- A track record of framing problems into learnable objectives.
- Hands-on dataset experience at scale: generation or collection, curation, automated quality assurance, coverage and bias analysis, and active learning to close edge cases systematically.
- Rigorous evaluation habits: baselines, ablations, metric design, and strong judgment for when a result is credible.
- Expert-level Python and PyTorch skills.
- Comfort with cloud training on GCP or AWS, plus Docker and experiment tracking tools.
- Professional experience with at least one of: multi-task model development and training; multi-modal model development and training; multi-view model development and training; self-supervised, outcome-based, or reward-driven learning.
Tech you’ll use
Python, PyTorch, GCP, AWS, Docker
Benefits
- Health, dental, and vision insurance
- Unlimited vacation
- Paid federal and state holidays
- 401(k) contributions of 5% of your salary
- Company-covered travel expenses
- Other perks to make your working life more fun, comfortable, and productive
Location: Charlestown, MA (onsite)
Salary: USD 170,100 - 227,400 per yearly
AI Tools Notice: We may use artificial intelligence (AI) tools to support parts of the hiring process, such as reviewing applications, analyzing resumes, or assessing responses and identifying potential inconsistencies or verification signals in application materials based on available information. These tools assist our recruitment team but do not replace human judgment. Final hiring decisions are ultimately made by humans. If you would like more information about how your data is processed, please contact us.
