AI Solution Engineer

AmeriLife • Remote, FL, NV

1d • $150,000 - $170,000 • Remote

About The Position

AmeriLife is a national leader in insurance and financial services, and we are standing up an enterprise AI capability from the ground up. The model is deliberately federated: a small, senior center owns the data platform, reusable AI services, and governance — while solution architects embedded in our Health and Wealth verticals find the highest-value work and build it alongside the business. This is one of the first of those embedded roles, and it is a builder’s job. You will spend most of your time engineering and shipping AI agents and LLM-powered services on Databricks and Azure — automating real workflows in contracting, commissions, and distribution operations where a national platform gives the economics real scale. The rest of your time draws on classic data science: the forecasting, propensity, and evaluation work that makes those solutions trustworthy and measurable.

Requirements

• 3+ years building AI or ML systems in production, including hands-on experience designing and shipping LLM-powered agents or multi-step AI workflows — not just consuming AI tools
• Practical fluency with at least one agent framework or SDK (Claude Agent SDK, LangGraph, LangChain, Databricks Mosaic AI Agent Framework, Semantic Kernel, or similar) and the ability to reason about why you chose it
• Tool and function calling: defining tools, wiring agents to internal APIs and data, and handling structured outputs reliably
• RAG and grounding in practice — chunking and retrieval strategy, vector search, semantic and hybrid retrieval, and knowing when retrieval is the wrong answer
• Prompt and context engineering as an engineering discipline: versioned, tested, and evaluated rather than hand-tuned
• Systematic AI evaluation — building eval sets, measuring quality and regression, and implementing guardrails for accuracy, safety, and cost
• Sound judgment on traditional ML versus generative AI versus deterministic automation, and the trade-offs of each
• Strong hands-on Databricks experience — notebooks, clusters, jobs and Workflows, and developing production-grade code rather than one-off analysis
• Advanced SQL and solid PySpark for large-scale transformation and feature engineering on a Lakehouse
• Unity Catalog for governance, lineage, and access control; Delta Lake and medallion architecture patterns
• MLflow for experiment tracking, model registry, and deployment
• Production experience on Microsoft Azure, including Azure OpenAI or Azure AI Foundry, and deploying services that other systems depend on
• Strong Python engineering practice: modular, tested, reviewable code with Git-based version control
• API design and integration — REST, authentication and secrets handling, and integrating with enterprise systems of record
• Containerization (Docker) and CI/CD for data and AI workloads
• Working understanding of cloud-native architecture, identity and RBAC, and data governance in a regulated environment
• Solid foundation in statistical modeling and machine learning, with the judgment to match the method to the business problem
• Experience building and validating supervised models on structured data (gradient boosting, regression, classification) and taking at least one to production
• Time-series forecasting experience, and comfort with hypothesis testing and rigorous model evaluation
• Comfort with imperfect real-world data — missing values, class imbalance, drift, and inconsistent source systems
• Demonstrated ability to work directly with non-technical business leaders — discovering opportunities, framing problems, and setting expectations honestly
• Full production ownership from problem definition through deployment, adoption, and iteration
• Experience leading delivery at the project or pod level: planning, sequencing, and accountability for an outcome
• Clear written and verbal communication, including the ability to explain a technical trade-off to an executive in a paragraph
• Bachelor’s or Master’s in Computer Science, Data Science, Engineering, Statistics, Applied Mathematics, or a related technical field. Equivalent experience with a strong portfolio of shipped work is equally welcome — show us what you have built.
• 6–10 years of combined software, data, or AI/ML engineering experience, with at least 2 years hands-on with LLM-based systems
• Must be authorized to work in the United States without sponsorship

Nice To Haves

• Hands-on work with Claude (Agent SDK, Claude Code, MCP) and/or building on Microsoft Copilot — Copilot Studio agents, M365 Copilot declarative agents and extensibility, Copilot connectors
• Building or consuming MCP servers to expose enterprise data and tools to agents
• Multi-agent orchestration, human-in-the-loop workflow design, or long-running agent state management
• Document intelligence and unstructured-data extraction at scale (forms, contracts, statements)
• LLM fine-tuning or adaptation, and a clear-eyed view of when it beats prompting or retrieval
• Databricks Mosaic AI — Agent Framework, Vector Search, Model Serving, AI Gateway, or Foundation Model APIs
• Delta Live Tables, Feature Store, Lakehouse Federation, or Databricks Asset Bundles
• Databricks certification (Data Engineer Professional, ML Engineer Professional, or Generative AI Engineer Associate)
• Azure Data Factory, Functions, API Management, Key Vault, Entra ID, Azure DevOps, or Logic Apps
• Infrastructure-as-code (Terraform, Bicep) and MLOps / LLMOps practice
• Azure certification (AI Engineer Associate, Data Scientist Associate, or Solutions Architect Expert)
• Clustering, anomaly detection, causal inference, uplift modeling, or Bayesian methods
• Experiment design and measurement in an operational (non-web) setting
• Optimization or simulation applied to a business process
• Insurance, financial services, healthcare, or another regulated industry — Medicare distribution, life and annuity, producer contracting, or commissions especially relevant
• Experience in a federated or multi-affiliate organization where influence matters more than authority
• Consulting, forward-deployed, or embedded-engineering background
• Track record of raising the technical bar around you — patterns, reviews, enablement, mentorship
• TypeScript a plus

Responsibilities

• Designing, building, evaluating, and shipping multi-step AI agents and LLM-powered services into production
• Finding and shaping high-value use cases with vertical leaders; reference architecture, reusable patterns, build-vs-buy input
• Forecasting, propensity and segmentation models, evaluation design, and the feature engineering behind both agents and models
• Build and ship AI agents
• Design, build, and deploy multi-step AI agents that complete real business workflows — retrieving from governed data, calling internal APIs and tools, making bounded decisions, and escalating to a human when they should.
• Engineer the unglamorous parts that make agents work: tool and function definitions, retrieval and grounding strategy, state and memory, orchestration, retries and failure handling, cost and latency management.
• Build evaluation into the build, not after it. Golden datasets, offline and online evals, regression suites, human-in-the-loop review, and guardrails you can point at when someone asks how you know it works.
• Instrument and operate what you ship. Tracing, monitoring, drift and quality alerting, and a clear owner for every production surface.
• Harvest reusable components into the shared services catalog so the next solution costs less than yours did.
• Architect solutions with the business
• Embed with your vertical’s leaders — operations, distribution, affiliate partners — observing the actual work rather than waiting on a written spec.
• Translate business problems into solution designs, including the honest version: what is automatable today, what needs process work first, and what is not worth building.
• Establish reference architectures and preferred patterns for your vertical, and contribute them back to the center.
• Bring judgment to build-versus-buy and to the question of when an agent is the right answer versus a model, a rule, or a fixed process.
• Apply data science where it moves the outcome
• Build and validate predictive models — forecasting, propensity, segmentation, anomaly detection — that inform planning or drive an automated decision.
• Engineer features and pipelines on the Lakehouse that serve both your models and your agents.
• Design the measurement. Baselines, holdouts, A/B and quasi-experimental designs, and a defensible read on whether the thing actually worked.
• Communicate results plainly to audiences that range from engineers to distribution executives.
• Deliver responsibly in a regulated business
• Document intended use, limitations, training-data assumptions, testing approach, and monitoring plan for every model and agent you put into production, and keep the model inventory current.
• Apply de-identification and least-privilege access as defaults when working with PHI, financial, or Medicare-related data.
• Flag fairness and unfair-discrimination risk on anything touching underwriting, rating, or pricing, and route it for actuarial and compliance review.
• Build for auditability — reproducible code, documented lineage and methodology, and recordkeeping that holds up under HIPAA, FINRA, SEC, CMS, and state insurance requirements.

Benefits

• PTO
• medical
• dental
• vision
• retirement savings
• disability insurance
• life insurance

What This Job Offers

• Job Type: Full-time
• Career Level: Senior