Technical Architect - Data Platform

Job Title: Technical Architect - Data Platform

Location: Remote

Job Type: FTE

Job Description

:Must-Have Skill

• s12+ years in data engineering and analytics; 3+ years as a Technical or Solution Architect on enterprise-scale, multi-workstream data program
• sHands-on expertise in Microsoft Fabric — OneLake, Fabric Data Factory, Lakehouses, Delta Lake, medallion architecture, and point-in-time reconstruction pattern
• sProven experience designing data quality and intake control frameworks — schema validation, SLA checks, reconciliation, exception triage queues, and quarantine/reject workflow
• sDeep experience designing canonical data models for complex transactional domains — insurance claims, healthcare, or financial services strongly preferre
• dHands-on experience with Azure AI Services — Azure Document Intelligence (Form Recognizer) or equivalent for unstructured/semi-structured document extraction integrated into pipeline
• sStrong Python and SQL — you review and write production-grade code, not just diagram
• sDemonstrated experience leading customer-facing technical engagements with full delivery accountability — you own the outcome, not just the architecture revie

w
Good to Ha

• veExperience with Azure Databricks for large-scale historical data processing or complex transformations alongside Fabr
• icPower BI embedded reporting or Fabric Real-Time Intelligence experien
• ceFamiliarity with medical stop loss insurance, TPA ecosystems, or healthcare claims data formats (837, 83
• 5)Microsoft Fabric Analytics Engineer (DP-600) or Azure AI Engineer Associate (AI-102) certificati

o

n Role Summa

ryWe are looking for a seasoned Technical Architect to own end-to-end solution architecture for a specialty medical stop loss insurance enterprise's CEO-prioritized claims data platform modernization. You will design and govern a production-ready Microsoft Fabric lakehouse — covering the full medallion stack from ingestion through curated Gold layer views — while serving as the primary technical authority in direct engagement with client leadership. This is a hands-on architecture role embedded in the delivery team, with a hard production deadline and real commercial stakes on every design decisio

n

. Why This Role Matte

rsThe customer processes 50,000+ claims annually across 250+ Third-Party Administrators using a fragmented, manual intake stack. The platform you architect will eliminate that friction, automate eligibility workflows, and become the data foundation on which all future AI/ML, BI, and operational capabilities are built. Phase 1 establishes a production-proven foundation on five priority TPAs — the architecture you get right here sets the pattern for the full TPA network in subsequent phase

s

. What You'll

DoArchitecture & Solution Desi

• gnDesign and govern the OneLake medallion architecture (Bronze / Silver / Gold) for claims, eligibility, and policy/contract data across five Phase 1 TP
• AsDefine the AI-assisted schema registry — versioned alias handling for ~10 TPA format variants each, with drift detection and forward-compatible schema evoluti
• onArchitect the canonical claims data model: standardized diagnosis/procedure/revenue codes, member/dependent/subscriber normalization, and full adjustment/void/reversal handli
• ngSpecify all six Gold layer curated 360 views — Insured/Claimant, Policy/Contract, Claim/Claim Line, Large Claimant, Clinical/Case Signal, and Underwriting/Renewal — and govern delivery to acceptance criter
• iaDesign the custom rule-based MDM claimant mastering service with probabilistic matching, human-in-the-loop review UI, and historical plus current-state covera

geData Platform & AI Integrati

• onGovern the Fabric Data Factory pipeline architecture for batch file intake (xlsx, csv, txt) with SLA monitoring, control totals, duplicate detection, and quarantine/reject workflo
• wsArchitect the Azure Document Intelligence integration for eligibility PDF and screenshot extraction, including human-in-the-loop validation workflow desi
• gnDefine Microsoft Purview adoption for data classification, lineage tracking, and 7-year PHI retention complian
• ceEmbed client's WinAIDM AI accelerators across all pipeline workstreams — Data Ingestion Agent, Data Quality Agent, Data Transformation Agent, and Reporter Age
• ntDrive the cross-validation strategy comparing new platform outputs against existing platform for accuracy and directional consisten

cyGovernance, Security & Complian

• ceDesign PHI access controls and data masking patterns enabling offshore development while meeting US-only PHI requiremen
• tsEstablish the four-role RBAC model (Admin, IT Support, Regular User, Read-Only) across the eligibility portal and all operational U
• IsDeliver architecture documentation, runbooks, and the Gold layer consumption specification for the client's downstream claim creation platfo

rmTechnical Leadersh

• ipLead sprint zero discovery — backlog definition, TPA file layout review, RACI, tooling access, and delivery cadence establishme
• ntConduct architecture decision reviews and set engineering standards across all workstreams and the full offshore te
• amMentor the offshore Technical Architect and Technical Lead on Fabric patterns, WinAIDM usage, and medallion implementation best practic

esClient Engageme

• ntServe as the primary technical voice in stakeholder meetings with client technology and business leadership — translating platform decisions into clear, decision-ready recommendatio
• nsPartner with the client's internal team to define Gold layer consumption spec and integration handoff criteria for the downstream claim creation platfo
• rmProactively surface trade-offs (build vs. configure, performance vs. cost, scope vs. timeline) and guide the client to the right decision — trusted advisor, not order-tak

er

What Success Looks Like (6–12 Mon

• ths)OneLake medallion architecture fully operational for five Phase 1 TPAs — Bronze immutable history, Silver canonical current state, Gold curated views consumed by the client's claim creation team before go-
• liveAI-assisted schema registry live and handling format drift across all Phase 1 TPA variants without manual interven
• tionIntake control layer operational with documented DQ rules per TPA, failed files routed to review queue with structured reason c
• odesWinAIDM accelerators deployed and driving measurable reduction in pipeline development and validation effort across the
• teamClient technology stakeholders describe client as a "trusted architecture guide" — proactive, decision-ready, and commercially ali

gned