Staff Machine Learning Engineer

BILL’s AI Product Engineering team builds machine learning systems that turn financial documents into accurate, actionable data. The Staff ML Engineer sets technical direction and develops production ML systems, leads LLM strategy and model evaluation, and mentors engineers across the organization.

Responsibilities:

• Own technical direction for high-impact ML initiatives end to end, from problem framing and research exploration through production deployment and measurement, across document understanding, field extraction, and model-serving infrastructure
• Stay hands-on: prototype, fine-tune, and ship models yourself; write production code alongside the engineers you mentor
• Lead our LLM strategy, including fine-tuning open-weight foundation models, designing evaluation harnesses, and making principled build-vs-buy decisions between self-hosted and frontier API models across accuracy, cost, latency, and privacy constraints
• Bring a research-oriented mindset to production problems: read and translate current literature into practice, design rigorous experiments and ablations, and know when a published technique will or won't transfer to our data distribution
• Design the data flywheel, including labeling pipelines, human-in-the-loop feedback, and drift monitoring, so our models improve continuously from production signals
• Set engineering standards for model evaluation, calibration, reproducibility, and responsible deployment; be the person others seek out for design reviews on anything ML-shaped
• Mentor and multiply: grow senior and mid-level ML engineers through code review, pairing, and technical guidance; influence roadmaps in partnership with product and platform leadership

Requirements:

• Requires a minimum of 8 years of related experience with a Bachelor's degree; or 6 years and a Master's degree; or a PhD with 3 years of experience; or equivalent experience
• 8+ years in software/ML engineering with a demonstrated track record at Staff level or equivalent scope, leading technically while remaining deeply hands-on with the modern ML/AI stack (PyTorch, distributed training, experiment tracking, feature and vector stores)
• Direct experience fine-tuning LLMs, including supervised fine-tuning, LoRA/QLoRA, preference optimization (DPO/RLHF), and structured-output training, and taking fine-tuned models to production at scale
• Production experience in at least one of: recommendation systems, personalization, or search/ranking; you have built systems that learn from user behavior at scale, handled feedback loops and position bias, and shipped models where offline metrics had to survive contact with online reality
• Strong data reasoning fundamentals: sampling and class imbalance, evaluation design and metric selection, confidence calibration, leakage detection, distribution shift, and the statistical judgment to know when an A/B result or benchmark number is real versus noise
• Research-oriented mindset with shipping discipline: you follow the literature, form hypotheses, and run disciplined experiments, but you measure success by what reaches production and moves business metrics
• Experience with document AI / intelligent document processing (OCR pipelines, layout-aware models such as LayoutLM/Donut, vision-language models)
• Publications, patents, or open-source contributions in ML
• Experience with privacy-constrained ML (on-prem/VPC model hosting, PII handling, data governance in fintech or healthcare)
• Experience designing multi-model routing or cascade architectures balancing cost, latency, and accuracy