Jason W. Stoudt

Signify Health (CVS Health)

As Lead Director, Engineering (Feb 2024–Aug 2025): led team-of-teams delivery across clinical and platform domains, authored ADRs, and began AI tooling adoption as Cursor champion. In August 2025, stepped off the management track deliberately — to go all-in on AI as an IC, prove the hypothesis, then build the platform for hundreds of others.

• Named by HCD CTO Amaresh Siva at HCD Townhall for AI SDLC program leadership; VP Shubha Chaudhuri publicly credited him as the architect of the AI SDLC program — “I want to make sure they get the credit they deserve”; the only IC reporting directly to a VP, with peers who are Exec Directors and AVPs.

Architecture & Technical Leadership

• Authored 10+ Architecture Decision Records spanning mobile platform (Swift vs. React Native, Nx module boundary governance), clinical systems (Build vs. Buy delta sync, CMS selection matrix), and developer tooling (ECC Portal strategy, Adobe vs. MixPanel, mobile drug database selection); issued architectural escalation during GitOps/Synapse transition (Nov 2025) that halted 12+ team migrations after identifying a systemic production risk — preventing a cross-org incident before it occurred.
• Served as Cursor AI champion for Atlas team (PCL-2403) during the Lead Director phase — running sprint-cadenced upskilling sessions for 6+ engineers and establishing the team’s early AI tooling foundation before the org-wide AI SDLC program launched.

Platform Engineering

• Built an end-to-end AI-native engineering platform — the first of its kind in the org — across five integrated layers: knowledge (Library-Agent, 1,372 repos indexed), framework (ai-native-dev-template, 100+ repos), specification (Relic, autonomous PR pipeline), evidence (AnchorKit/EFD, executable specs), and governance (Hallucination Ward, quorum voting + CHAOS testing). Each layer independently useful; together, a complete answer to enterprise-scale AI engineering.
• Designed the AI-native development framework propagated to 100+ repositories; led a framework simplification audit cutting LLM context overhead by 60% (2,932 → 1,180 always-apply lines, 4,948 lines removed) — reducing cost and latency for every consuming team.
• Created Evidence-First Development (AnchorKit): replaces 12–20 drift-prone markdown artifacts per feature with 1–3 executable files (Gherkin .feature + optional eval suite + optional ADR), now adopted as engineering standard; also built Repo-Sage, which mines PR review history to auto-generate CLAUDE.md team conventions.
• Built the org’s first fully autonomous software delivery pipeline (Relic, 247 PRs): non-developers file a GitHub Issue; AI agents run the full specify→plan→implement cycle and deliver a Draft PR — no engineering involvement required. Not a productivity improvement: a permanent change in who can initiate software delivery. Also built Library-Agent (org knowledge API, 1,372 repos indexed, 246 PRs) and Member-Agent (voice-enabled clinician training system, Gang-of-15 reviewed).

Governance

• Designed and operates an autonomous AI governance platform (Hallucination Ward / jAIson Bourne agent): formal proposal process with quorum-based voting and staggered veto windows, adversarial stress testing via 22+ CHAOS tabletop exercises, and 400+ governed sessions generating ~513 autonomous GitHub PRs — the first AI governance system of its kind in the organization — governance is the constraint that stalls most enterprise AI programs; this system solved it: 513 autonomous PRs delivered under formal oversight, with zero production incidents attributable to autonomous AI.

Enterprise Program

• Founded the company-wide AI SDLC program — 3–5× velocity improvement (April 2026: 360 PRs in one month, 70% of the prior 18-month baseline); 11 teams onboarded in 2 days vs. prior 1–3 months; live demo: 7,500 lines of production-ready code across 105 files at ~$235 LLM cost; hosts the 2×/weekly Forum (80–100+ attendees, US/Ireland) and drives the weekly Executive SteerCo; facilitated 3 immersion workshops (Dallas ×2, Galway); authored the AI SDLC Autonomy Framework (L0–L5 governance, Tier 1–4 classification).

Clinical Delivery

• Authored 43 technical specifications for a greenfield Clinical Referral Management System in a single day — FHIR R4, Kafka/NServiceBus, .NET 10, React 19, 8-state state machine, PHI/HIPAA gate; led Gang-of-N AI reviews: G7 (64 architecture issues), G15 (5 P0 fixed), G30 (4,547 lines reviewed).

Operations

• Owned 4 production incidents: billing-blocking defect across 13,000 evaluations (direct revenue impact, surfaced over a weekend before business hours); medication pre-population failure (2,061 providers, LaunchDarkly rollback); duplicate-evaluation slow leak (SQL mitigation + RCA); clinical app outage detected before 8am.

The transformation in aggregate: AI-native framework deployed to 100+ repositories; org knowledge API indexing 1,372 repositories; 300+ engineers reached through Forum and immersion workshops; first formal AI governance system in the org; 513 autonomous PRs governed; 11 teams onboarded in 2 days vs. prior 1–3 months each. From a single IC bet in August 2025 to company-wide transformation recognized at the CEO level — in 10 months. His 1,234 PRs at 4.3× are the least interesting measure of impact; the larger story is the perpetual leverage of the platform, used daily by hundreds of engineers.

Osmosis from Elsevier

• Accountable for five software development teams, each comprising an Engineering Lead, 2–4 developers, and a Quality Engineer.
• Led migration of the front-end to React; implemented i18n for Spanish, broadening accessibility to a new market segment.
• Delivered full CI-CD in GitHub Actions with end-to-end smoke tests and regression suites, ensuring high-quality, consistent releases.
• Streamlined authentication via Auth0 SSO (Google & Facebook), built a new cross-product shared data pipeline, and provisioned an anonymized demo environment.
• Absorbed and integrated the Business Intelligence team into the engineering org, promoting cross-functional collaboration.
• Removed legacy Fibers JavaScript library; introduced GitHub Copilot, New Relic, Checkmarx, and SonarQube to sharpen quality and security posture.

Elsevier

• Managed back-end component teams for ClinicalKey — Elsevier’s clinical knowledge platform serving 130M+ healthcare professionals globally — 9 direct reports, 14 total headcount.
• Migrated 15 back-end Java repositories from Maven to Gradle, dramatically reducing local and CI-CD build times.
• Rolled out SonarQube static analysis across all ClinicalKey repositories (Java & JavaScript).
• Restructured development squads around component ownership, improving efficiency, shared knowledge, and team autonomy.
• Introduced ephemeral environments, eliminating dual long-lived branches and the overhead that came with them.

Elsevier

Lead engineer on Elsa — Elsevier’s global end-to-end content creation platform supporting ~500K article submissions per year across 2,500+ journals. Directed four teams with 20+ developers across on-shore and off-shore.

• Designed and executed a migration from Backbone.js to React, using Lerna to manage the front-end monorepo.
• Consolidated ~100 back-end Java repositories from Maven into a single Gradle monorepo, significantly cutting build times.
• Introduced SonarQube scanning in CI-CD for long-lived branches and pull request builds.

L3 Technologies – Communications Systems East

• Chief Software Architect for PTES — a cloud-based mission planning and key management system on DISA infrastructure; designed microservices architecture on Red Hat OpenShift PaaS running in Docker containers, one of the earliest DoD cloud deployments of its type.
• Chief Software Architect for automated embedded build system (C++/C): replaced a fully manual process with Gradle-based dependency management and automated builds — projected savings of hundreds of thousands to millions of dollars annually across the program portfolio.
• Lead Software Architect for PKMA across 3 Scrum teams (18 development sprints); pioneered Agile Scrum adoption at L3 — team recognized as benchmark for 10 other Scrum teams; led transition from ACTS KMA to PKMA with zero severe defects.
• Chief Software Architect for PKL key fill device (Qt 5.7); Software Design Lead on Stingray (5 engineers, Java/Spring/PostgreSQL) and ACTS KMA Key Translation Element (10 engineers, C/PL/SQL on UNIX); served as integration lead and on-site customer liaison across 5+ programs.
• Held DoD Secret (2002) and DoD Top Secret / SCI Counterintelligence (2011) clearances (both inactive since 2018); annually ranked among top engineers; received personal recognition awards on every program.

Signify Health (CVS Health)

2×/week · 80–100+ attendees per session · AM and PM slots (US and Ireland time zones)

Signify Health (CVS Health)

Weekly strategy sessions with VP, Lead Director, and Exec Director stakeholders

Signify Health / CVS Health — Dallas, TX

12+ engineers and executives including directors and VP

Signify Health — Galway, Ireland (Reformers Team)

Named “indispensable” for the rollout by VP Casey Lucas