Note: The job is a remote job and is open to candidates in USA. Jobber provides technology that helps small home service businesses manage quoting, scheduling, invoicing, payments, and customer experiences. The Staff Data Scientist, Applied Machine Learning will own automated decisioning models and production ML systems, including SignalGraph, retrieval-augmented generation, real-time serving, monitoring, evaluation, and retraining. The role also sets technical standards, mentors scientists, and partners with senior leaders and cross-functional teams to apply machine learning to business decisions.
Responsibilities
- Continue building, improving, and maintaining SignalGraph: evolve metric-family contracts, keep segment and causal-edge catalogs trustworthy, operate the Neo4j and series-refresh pipeline, and strengthen investigation diagnostics so teams can trace metric changes to root causes and make faster decisions from governed evidence
- Design, build, and evaluate retrieval-augmented generation (RAG) systems on top of that graph. You’ll own retrieval quality, context management, and the evaluation harness that proves the system is actually right, not just fluent
- Own production ML end-to-end: training pipelines, real-time serving, monitoring, drift detection, and retraining. When a model is live, its uptime, latency, and accuracy are yours
- Establish systematic evaluation and regression testing as a standard for the team — so model and LLM system quality is measured and defended over time rather than assessed once at launch
- Set the technical bar and multiply the team: drive MLOps and ML engineering standards, shape our feature store and platform roadmap, review peers' work, and mentor other scientists on graph and deep learning methods
- Partner directly with senior leadership. Much of what you build will be used by Senior Leaders, Customer Analytics, Business Intelligence, and Product to make decisions against Jobber's North Star goals. You'll present your own work, defend your assumptions, and help cross-functional partners reframe how they approach a problem
- Stay current, and bring it back. We expect this role to be the most informed person at Jobber on developments in AI/ML methodologically, not just as a consumer of tools, and to translate that into shipped capability
Skills
- Production ML experience, end-to-end. You've trained, deployed, served, and maintained models that acted on real users (not just scored offline) You've learned the hard lessons about retraining, drift, and technical debt that come with it
- A strong statistics foundation. You can reason about bias and variance, justify a loss function, and tell signal from noise without hand-waving
- Expert SQL and production-grade Python skills
- Depth in modern ML methods: Deep learning and neural architectures, transformers/BERT-family models, RNN/CNN, ranking, and representation learning, with hands-on experience applying LLMs in production systems, including RAG and context management
- Experience designing for the real cost of being wrong. You've built or tuned custom and asymmetric loss functions where a miss in one direction is far more expensive than the other, and you can explain the business reasoning behind the choice
- Experience with large data in production environments and the platforms that support it
- ML/AI platforms such as Snowflake, orchestration with Apache Airflow, and cloud infrastructure (AWS strongly preferred; GCP or Azure equivalent considered)
- Strong communication and stakeholder alignment skills. This team's work only lands if everyone shares the same understanding of what the data means. You can take a room of technical and non-technical partners from confusion to alignment, present uncertainty honestly, and influence without authority
- Ownership of quality over shipping speed alone. You want to know •why• something works and how to make it better. You don't accept a model's current performance as its ceiling, and you don't outsource your understanding of your own code to an AI assistant
- AWS strongly preferred; GCP or Azure equivalent considered
- Graph experience: Graph theory, graph neural networks, knowledge graphs, or graph databases such as Neo4j. Rare, and the single most valuable thing you could bring to this role
- A software engineering background: Services and model serving (REST/gRPC), Docker/Kubernetes, CI/CD, feature stores
- Experience with Snowflake, including Snowpark and Snowpark Container Services, or a comparable path from warehouse to deployed model
- Experience building LLM evaluation infrastructure, safety layers, or LLMOps for customer-facing AI
- Exposure to risk, fraud, or fintech modelling, or to recommendation and ranking systems at scale
Benefits
- Equity rewards
- Annual stipends for health and wellness
- Retirement savings matching
- An extended health package with fully paid premiums for body and mind
- A dedicated talent development program that includes career coaching and opportunities for career development
- An extended health benefits package with fully paid premiums for both body and mind
- Matching in RRSP, TFSA or FHSA
- Stock options
- Access to coaching, learning, and leadership programs to help you grow your career, reach your goals, and unlock your full potential
- Hybrid work arrangement
Company Overview
Company H1B Sponsorship