AI Senior Support Engineer

AI Senior Support Engineer

Hours: Aligned to Chicago time (CT) · Engagement: 80 hrs/month, full-stack + data stack remit

About the Role

Altoros is staffing a Senior Support Engineer for a client engagement supporting an analytics platform. This is a full-stack and data-stack role: the engineer owns the platform shell, the ingestion and semantic layer, and a bounded amount of embedded analytics component upkeep. The role is commercial-model first: month one is an observe and define phase (baseline setting, onboarding), with the engagement shifting toward outcome-based delivery from month two (issue resolution time, defect reduction against baseline, availability targets).

AI-augmented delivery is central to this role and one of its most important elements. Working AI-first with Claude Code is how a single engineer credibly covers this full remit. Altoros builds its delivery on Anthropic's professional courses and certification, and the engineer uses Claude Code across the full range of work — maintenance, bug-fixing, and data work, not just new development — operating inside the client's own Claude Code / AI-tooling accounts.

Scope & Responsibilities

Platform Shell

Maintain and extend authentication and SSO integration

Own navigation and application shell components (modern front-end framework, e.g., React / TypeScript)

Manage tenant and user management, including multi-tenancy considerations

Data & Semantic Stack

Build and maintain data ingestion pipelines

Operate and extend orchestration workflows in Dagster

Develop and maintain transformation logic in dbt and SQL

Work with the data warehouse on Google Cloud Platform (GCP), primarily BigQuery

Maintain the semantic layer in Cube, including metric definitions and data modeling

Embedded Analytics (light, bounded)

Customize and update Embeddable component files pulled into the client's repo

Maintain theming and keep the Embeddable SDK current

Note: Embeddable itself handles the builder, embed serving/rendering, security tokens, and multi-tenancy: this is a maintenance layer, not a build-from-scratch effort

Delivery & Documentation

Maintain centralized documentation in Confluence, including DBML/database diagrams

Capture ongoing knowledge for handover and continuity purposes

Work to defined outcome targets from month two: P1 issue resolution/mitigation within one business day, defect reduction against an agreed baseline, and business-hours availability once the client is live

Use spare capacity (when live issues don't consume the monthly band) on preventative maintenance, hardening, and onboarding new data sources/integrations

Required Skills & Experience

AI-first delivery (core requirement): hands-on with Claude Code (or similar) / AI-assisted engineering across the full development lifecycle; Anthropic's professional courses and certification are a strong plus (or readiness to complete them)

Full-stack development experience, including a modern front-end framework (e.g. React / TypeScript), authentication/SSO implementation, and multi-tenant application architecture

Hands-on experience with Dagster for orchestration (or similar tools)

Strong DBT and SQL experience for data transformation

Experience with BigQuery and the Google Cloud Platform (GCP) data stack

Experience with Cube or a comparable semantic-layer / metrics-layer tool

Familiarity with embedded analytics tooling (Embeddable or similar), component customization, theming, SDK integration

Comfortable working independently and engaging directly with client stakeholders

Strong documentation discipline: Confluence, DBML/ER diagrams

Available to work core hours aligned to Chicago time (CT)

Nice to Have

Background supporting analytics/BI platforms for enterprise or sports/media clients

Experience setting SLA style targets (resolution time, availability) and reporting against them

Engagement Details

80 hours/month, full remit across platform shell, data/semantic stack, and Embeddable upkeep

Backup coverage required for continuity during absences: candidate should be able to hand off context cleanly