The company builds agentic intelligence services for clients, combining expert people with AI agents. It was named to the 2026 Inc. 5000 Top 40 fastest-growing companies in AI and Data and ranked 723 overall in the US.
The team values independent judgment, high standards and speed, and it judges work by its impact rather than its activity.
Ownership is substantial, leaders are directly reachable, and scope grows as quickly as you can show you can handle it. Advancement depends on contribution, not tenure or office politics.
The role suits high-potential people who learn quickly, take feedback well and want work larger than their current title. Agents support your judgment and do not replace it or lower the bar.
Responsibilities
Own the intelligence layer of client engagements: how the system reasons, how its correctness is verified and how results are proven.
Work inside client data alongside engineers and consultants to decide whether an agentic workflow can be trusted in production.
Decide which tasks need a model and which need a rule, and set how good is good enough for each decision.
Set up human review for outputs that need it, and detect quality drops after release.
Build models and evaluation systems yourself, and define the measures used to judge the outcome.
Tell executives directly when the data cannot support the claim they want to make.
Develop evaluation methods for production agents that will be reused across healthcare, financial services and marketing.
Embed with client teams to find, profile and assess data quality and lineage, and state what the data can and cannot support.
Turn business outcomes into measurable problems, set baselines and design measurement, using holdouts and experiments where possible.
Choose the method for each workflow step (LLM reasoning, classical machine learning, statistical models or rules) based on accuracy, cost, latency and risk.
Build evaluation frameworks for agentic systems, including reference datasets from real client cases, automated graders, calibrated human review and regression suites for model, prompt and tool changes.
Develop and deploy models for scoring, forecasting, classification, anomaly and signal detection, retrieval and ranking where they fit.
Design risk-tiered review, including confidence thresholds and routing logic that decide which outputs proceed and which go to a person.
Test for bias and uneven performance across affected populations and treat findings as design constraints.
Monitor quality, drift, exceptions and cost in production and define triggers for reassessment.
Work with engineers to move your work into production using sound MLOps practice.
Measure results against the baseline and report them with their limits, including limits on attribution.
Explain methods, results and uncertainty to client operators and executives in terms they can act on.
Contribute evaluation harnesses, measurement templates and model patterns to the shared foundation and mentor data scientists and engineers on your team.
Requirements
A bachelor's degree in a quantitative field or equivalent experience, plus 5+ years of applied data science or machine learning work, including systems you built that run in production.
Strong Python and SQL programming skills.
Deep knowledge of statistics, experimentation and causal inference, with the judgment to choose the right method for a question.
Hands-on experience building and evaluating LLM applications.
Ability to present complex analysis to technical and non-technical audiences, starting with the decision.
Experience working directly with clients or business stakeholders.
Willingness to travel to client sites as engagements require.
Preferred: an advanced degree (master's or PhD) in statistics, computer science, mathematics or another quantitative field.
Preferred: experience with marketing measurement, attribution or customer journey analytics.
Preferred: experience with healthcare or life sciences data, including FHIR, claims or clinical data.
Preferred: experience with MLOps tooling, cloud machine learning platforms and model monitoring in production.
Preferred: experience with responsible AI practice, including fairness assessment and model documentation.
Candidates who do not match every item are still encouraged to apply and explain what they would bring.
Conditions
Client site travel is required as engagements require.