Senior Machine Learning (ML) Engineer

We are building a personal intelligence for Mac that learns your patterns over time and helps you see, protect, and steer your attention. Raw data stays on-device. We're a small, venture-backed team (BetaWorks, True Ventures, Slack Fund, RRE) building from 0 to 1 in New York.

We're hiring an ML engineer to build the model training and inference systems behind a Mac-native personal intelligence product. You'll work with high-frequency behavioral signals collected on-device, build models that learn and personalize over time, and ship pipelines that run reliably on Apple Silicon hardware. 

This is real ML — not LLM wrappers. 

You'll train classical and lightweight temporal models on behavioral data, build the pipelines that feed them, and own the systems that keep them running in production. 

You'll have access to two NVIDIA DGX Sparks for training, plus generous compute credits for experimentation. 

We're an AI-forward engineering team. You'll be expected to use frontier coding tools — Claude Code, Codex, or similar — as a core part of how you build. 

We optimize for compound leverage: systems that test, monitor, and improve themselves over time with minimal manual upkeep.

Small, collaborative founding team — minimal bureaucracy, high ownership. AI-forward development environment focused on speed and self-maintaining systems.

• Interview process that respects people and their time
• Professional and open IT community
• Internal meet-ups and resources for knowledge sharing
• Time for recovery and relaxation
• Bright online and offline events
• Opportunity to become part of our internal volunteer community

Why This Role

• Real ML — temporal and classical models on behavioral data.
• Real data and compute — high-fidelity signals and DGX training access.
• High ownership in a small team.
• Mission-driven product helping users reclaim attention.

What We're Not Looking For

• Primarily LLM fine-tuning, prompt engineering, or chatbot-focused ML experience.

• Notebook-only data scientists without production ML ownership.

• Someone who needs detailed step-by-step direction in ambiguous environments.

• Technically strong but poor communicator — or great communicator who struggles to ship. 

  Responsibilities

• Train and evaluate classical ML models — gradient boosted models, logistic regression, lightweight time series models (LSTM, temporal CNN).

• Build model training pipelines — make the model training process repeatable once proven.

• Optimize for on-device constraints: low-latency inference while respecting battery and thermal budgets.

• Build data pipelines for ingesting, processing, and structuring behavioral data.

• Implement evaluation and monitoring: confidence tracking, retraining triggers, model versioning.

• Collaborate with the founding team to translate product requirements into model improvements.

• 5+ years in ML engineering or closely related production systems role.
• Strong Python skills for ML pipelines and data processing.
• Classical ML expertise: gradient boosted models, time series analysis, practical model selection.
• AI-forward engineering mindset using modern coding tools.
• Non-LLM focus: behavioral models (pattern recognition, state classification, anomaly detection).
• Experience with synthetic data (nice to have).
• On-device or edge ML experience (nice to have).
• Production discipline: monitoring, versioning, reproducibility.
• Familiarity with Apple/Swift ecosystem (nice to have).