The posting
The Opportunity
As a Senior Data Scientist on the Design team, you will play a pivotal role in transforming how some of the world's most critical infrastructure companies operate. The experiences you shape for both humans and agents will directly influence how complex projects are planned, deployed, and managed at scale.
We are rethinking what excellent UX means in an agentic world. When the interface is a model, the quality of the experience is no longer carried primarily by pixels and flows — it is carried by behavior. What the agent retrieves, what it decides, what it says, when it refuses, and how reliably it does all of that across thousands of real situations. Those are measurement problems, experimentation problems, and evaluation problems. They are your problems.
This role is ideal for someone who combines statistical rigor with a strong builder mentality — someone energized by ambiguity, fluent in model behavior, and obsessed with what "good" actually means and how to prove it. You will set the standard for how agent experiences are designed, instrumented, and held to a quality bar at Sitetracker — hands-on, technically deep, and relentlessly connected to customer outcomes.
You will be embedded in a scrum team — designing, measuring, and delivering alongside Engineering and Design on a day-to-day basis. You will own day-to-day decisions for your experience area, mentor others on rigor and method, and ship to production. For someone who wants to build, measure, and elevate the craft around them, this is the role.
What You'll Do
Own Your Domain and Deliver High-Impact Solutions
You will take ownership of day-to-day decisions for your experience area. You will take priorities from leadership and define solutions independently — making informed tradeoff decisions across user needs, agent behavior, technical constraints, and business goals. You will define the leading indicators that tell you whether an experience is working, scope work, prioritize effectively, and own the consequences of your decisions. You will ship fast without sacrificing quality, knowing when to ship on signal and when you need more.
Study the Market and the Technology
You will stay close to where product and AI capabilities are heading, not just what Sitetracker can do today. You will maintain running awareness of the competitive landscape across multiple verticals, identify gaps and opportunities, and produce analysis that connects directly to product and behavior decisions. You will develop a deep understanding of the critical infrastructure industries Sitetracker serves — telecommunications, EV charging, renewables, and utilities — and the workflows professionals in those industries rely on daily, because you cannot evaluate an agent's answers without understanding what a right answer looks like in the field. You will speak the technical language — fluent enough in model behavior, retrieval, platform capabilities, and enterprise architecture to hold your own as a peer in deep technical discussions, not a passenger.
Investigate Complex Problems
You will investigate complex, cross-team problems, synthesizing signal from logs, traces, evals, and customer behavior to separate root causes from symptoms. You will own investigations end to end — instrumenting what isn't measured yet, designing the analysis, and moving from finding to action quickly. You will work directly with customers — including those with complex needs or strong opinions — running pilots, mining how they actually use the product, and turning behavioral patterns into design decisions. You will show up to customer calls thoroughly prepared with account context and the data to back it.
Design the Full Agentic Experience
You will design comprehensive solutions that think in systems and behavior, not screens. You will define what the agent should do across the full distribution of inputs it will see — not just the happy path — and design the context architecture, decision paths, and communication patterns that get it there. You will map where the agent surfaces across the product, define the behavioral contract for each surface, and account for how changes ripple across the experience. Sitetracker is a Salesforce ISV — our web experience is built on SLDS-based components, while our mobile experience is purpose-built outside of SLDS. You will design for the full range of our users across both — from field technicians on mobile to program managers running complex dashboards — and for the agent that now sits between them and the work.
Build and Ship
The discipline is changing. The most effective people on a product team are no longer handing off static artifacts and waiting for someone else to bring their ideas to life. They are shipping code, building AI graphs, measuring the quality of their solutions, and delivering outcomes directly. Sitetracker is at the front edge of this evolution, and this role will help define what it looks like.
You are a builder. A notebook is one tool in the kit, not the destination. You will ship code to production using AI-assisted development tools like Claude Code, Cursor, and Scout itself, build and refine the data pipelines and eval harnesses that keep quality honest, and verify implementations directly. You will build flexible prototypes and instrumentation, loop engineers in for review, and write code that works within the existing codebase. You will be a trusted partner to your engineering counterparts, actively engaging in engineering Slack channels and contributing to technical discussions.
Architect and Evaluate Scout AI Experiences
You will architect Scout graphs and design sophisticated AI interactions. You will be fluent in the technical language — able to reason about graph logic, context retrieval, grounding, and AI behavior independently. You will design Scout's decision paths, context architecture, and communication patterns, and define how Scout surfaces throughout the broader product. Most of all, you will own quality: you will write comprehensive evals across multiple dimensions — correctness, grounding, safety, refusal, latency, tone — define what success means quantitatively, and track it over time in Watch. You will design the offline and online evaluation strategy for your area, build the datasets that power it, and stay current with the broader AI landscape so your eval design keeps pace with what the models can and can't do.
Validate and Measure Quality
This is the center of the role. You will strategically choose when and how to validate, and you will bring real method to it. You will design experiments — A/B tests, holdouts, shadow evaluations — to de-risk before customers are involved and to learn from real behavior once they are. You will distinguish a real effect from noise, size experiments correctly, and resist shipping on anecdote. You will run usability and behavior reviews for complex agent features across verticals, use AI-assisted workflows to build high-fidelity prototypes that test real Scout behavior, and run customer betas and previews — including with customers who have more complex needs or stronger opinions. You will establish metrics connected to commercial outcomes and consider your work done when it's succeeding, not when it ships.
Drive Adoption and Outcomes
Your work is not done when a feature ships. It's done when it succeeds. You will track whether what you built is adopted and whether it drives the commercial outcomes it was designed to achieve. You will build the dashboards and analysis that make adoption and outcomes legible to the rest of the org, and you will apply what you learn to iterate, adjust, and improve.
Mentor and Elevate the Team
You will raise the rigor bar around you. You will review others' work — evals, experiments, analyses, and agent designs — and provide substantive, actionable feedback that improves output. You will teach the team how to define quality, design an eval, and read an experiment honestly. You will model the "build in public" culture by sharing work early and often across broad channels, inviting critique, and showing how work evolves based on input. You may manage a dedicated intern, setting clear expectations and ensuring they do meaningful work.
The Skills You'll Have
Act with Agency
- Thrive in fast-moving environments and deliver high-impact outcomes at pace
- Independently identify ambiguous challenges and take ownership of driving them to resolution
- Champion new ideas, influence stakeholders, and rally support around strategic initiatives
- Identify workflows where AI tools and emerging technologies accelerate delivery, and make building the default way you work
- Measurement, Experimentation, and Evaluation
- Define what "good" means for an agentic experience and turn it into metrics, evals, and datasets that hold up under scrutiny
- Design and analyze experiments — A/B tests, holdouts, shadow runs — with correct sizing, clean reads, and honest conclusions
- Build offline and online evaluation systems for LLM behavior across correctness, grounding, safety, tone, and latency
- Apply statistical and ML fluency to real, messy production behavior — distinguishing signal from noise and root cause from symptom
- Agentic UX and Behavior Design
- Turn complex enterprise workflows into agent experiences people can actually use without training
- Design context architecture, decision paths, and communication patterns, and reason about how they shape behavior across the full input distribution
- Work directly with customers — running pilots, mining usage, and iterating based on how they actually behave, including customers with complex needs or in delicate situations
- Make tradeoff decisions across user needs, agent behavior, technical constraints, and business goals, and connect those decisions to commercial outcomes
Technical Fluency and Building
- Open PRs that ship work to production, build functional prototypes and pipelines, and use AI-assisted development tools as a primary working medium
- Build instrumentation and eval harnesses that are reusable and work within the existing codebase
- Speak the technical language as a peer — comfortable in discussions about platform architecture, retrieval, grounding, and engineering tradeoffs
- Architect Scout graphs and reason about graph logic, context retrieval, and AI behavior independently
Competitive & Market Awareness
- Maintain running awareness of the competitive landscape across multiple verticals
- Identify gaps and opportunities and produce analysis that connects to product and behavior decisions
- Track market trends, adjacent products, and emerging tools, connecting what's happening in the market to what Sitetracker should be doing
- Regularly contribute to competition and industry Slack channels with substantive observations
Product & Industry Knowledge
- Deep product knowledge across your area and adjacent areas
- Understand how customers incorporate Scout into industry-specific workflows, and what a correct answer looks like in each
- Seek out customer conversations proactively and use what you learn to inform decisions
- Can demo your area in depth and give a solid cross-product demo
Presentation & Communication
- Present compellingly to senior stakeholders and bring people along through the process
- Model the "build in public" culture by sharing work early and often, inviting critique, and showing how work evolves based on input
- Translate complex technical and quantitative concepts for diverse audiences
Within 90 Days, You'll:
- Collaborate regularly with product management, design, and engineering peers, and have met key stakeholders across the departments you work with
- Be on the phone with customers, learning their workflows, pain points, and how they use Sitetracker and Scout. Know the personas and subverticals
- Deliver work to your scrum team — pushing code, building instrumentation and evals, and demonstrating the builder standard
- Be able to confidently demo Sitetracker's core product in your area and give a solid cross-product demo
- Stand up baseline evals and quality tracking for your area in Watch, and establish what you don't yet measure
- Begin mentoring others and providing substantive feedback in reviews
Within 180 Days, You'll:
- Lead the successful delivery of high-impact agent experiences across complex enterprise workflows
- Own day-to-day decisions for your experience area independently, establishing leading indicators and making tradeoff decisions confidently
- Architect Scout graphs and design sophisticated AI interactions, owning a comprehensive eval suite and tracking quality over time
- Design and run experiments on real agent behavior, and run customer betas and previews independently — including with customers who have more complex needs or stronger opinions
- Actively mentor others and raise the bar through consistent, high-quality reviews of evals, experiments, and agent designs
- Contribute meaningfully to shared evaluation infrastructure and establish higher standards for quality measurement in your domain
Within 365 Days, You'll:
- Point to product successes you have driven that have directly contributed to commercial outcomes
- Be recognized as an expert in your domain — people across the org come to you when they need to understand your experience area or how its quality is measured
- Ship code confidently to production, build the instrumentation and evals that keep quality honest, and operate as a true peer with engineering
- Be trusted with the most complex customer situations and maintain relationships with key customers that you built and maintain on your own
- Shape direction across multiple projects with well-reasoned, evidence-backed positions that influence the roadmap
- Set the pace for delivery in your domain and make others faster by reducing ambiguity early
