SAP, enterprise systems & program leadership
Large-scale technical program management, integration, enterprise operations, production support, and technical leadership.
Resume-backed engineering portfolio
Roger Murphy
Senior SAP, government cloud, technical program, and AI platform leader
I lead and deliver large-scale enterprise programs, then operate what was promised: 20+ years of SAP delivery, enterprise integration, and technical program leadership; regulated government-cloud operations since 2021; and $82M in Premium Engagement revenue driven scope-to-close through presales technical solutioning. I still build hands-on — the current proof is oshal / Open Swarm, a governed multi-agent platform with real applications running on it, covered further down this page.
20+years of SAP delivery, technical program leadership, and enterprise operations
5years of regulated government-cloud and secure SAP operations (2021–present)
$82Min Premium Engagement revenue driven scope-to-close, including 25 net-new contracts
1agentic framework — oshal / Open Swarm, with real applications running on it
Career at a glance
Twenty years of delivery, five in government cloud, and a hands-on approach to leadership.
The full story is in the resume; the detailed companies and project addendum carries the deeper ledger.
20+ years
Enterprise SAP & program leadership
Delivered and supported SAP-centered enterprise environments where reliability, integration, and business continuity matter more than novelty — including safeguarding a $500M reimplementation to a zero-defect go-live. Translated technical constraints into plans, estimates, and demos non-specialists could evaluate.
More on About -> Since 2021Government cloud & regulated operations
Operated where identity, data boundaries, change control, and documentation are part of the engineering surface: secure SAP cloud operations, escalation governance, incident response, and evidence-driven decisions.
More on About -> $82M closedPresales, solutioning & capture
Drove scope-to-close execution for $82M in Premium Engagement revenue — discovery, value-case development, service scoping, delivery validation, quote-to-order governance, and margin discipline — with 25 net-new contracts and 60+ customer workshops. A winning solution has to be both persuasive and executable.
More on About -> Current buildoshal / Open Swarm
A governed multi-agent platform built and operated solo: controller, bot-node mesh, workflow engine, brokered connectors, tenant isolation, and a per-call cost ledger — with real applications loading from manifests.
More on About ->Core strengths
What I bring to a team.
Government cloud & regulated operations
Security-aware operations, access boundaries, evidence, incident response, and platform discipline.
Presales & capture engineering
Solution architecture, technical volumes, estimates, demos, transition thinking, and delivery credibility.
AI platform engineering
Agentic systems, bot orchestration, workflow automation, connector brokering, RAG, local and hosted models, and governed tool execution.
Cloud, DevOps & operations
Docker, Kubernetes, Terraform, OIDC/Keycloak, incident workflows, observability, and automation.
Full-stack delivery
TypeScript, Node.js, Python, application surfaces, APIs, Postgres, Redis Streams, and deployment.
The current build
oshal / Open Swarm: the platform I build and operate.
One well-documented exhibit alongside the career above. It applies enterprise discipline to AI systems — tenant boundaries, row-level security, review gates, per-agent tool policy, traceable work, and cost capture. The controller routes work, bot nodes run the model, connectors broker outside access, and applications load from manifests, not forks.
38applications on one runtime, each loaded from a manifest
306audited connectors, conformance-checked by a nightly evidence run
843unit tests green on every commit, CI-gated
79architecture decision records documenting the build
Right tool per job
Each bot picks its harness, provider, and model — swap any of them without touching the workflow.
Accountable by design
Tickets, review gates, per-agent tool auth, and cost captured once, centrally — not a black box.
Safe to share
Postgres row-level security and per-user storage wall off tenants and users inside one runtime — enforced by default, proven with a live two-user isolation test.
Cost you can optimize
Every call’s tokens and cost land in a central ledger per bot. Captured work replays across models and providers and is graded — so the same job finds its cheapest configuration that holds quality.
TypeScriptRedis StreamsPostgres + RLSChromaDB RAGDocker · KubernetesOIDC / KeycloakConnector brokerLocal LLMs
Beta. A guest session is a read-only look at the live cockpit; signing in with Google creates your own workspace on the spot. The trading account is paper (simulated).
One-page view
The whole system on one page.
Two views of the same build: the runtime architecture — applications, workflow engine, connectors, isolation, and deployment — and the application fleet radiating from one framework.
On the platform
Applications running on oshal.
The flagships — an autonomous trading desk, Intelligent Career Search, and the Little Monsters study app — plus the wider fleet. Each links to its own page.
Live · paper account
Intelligent Trading
A swarm runs a paper book end to end: signals, risk, fills, and a daily AI-generated video recap. Every trade is tied to the signal that justified it, with a live account board on the page.
See the record & video -> Intelligent appIntelligent Career Search
Turns scattered experience — files, notes, and spoken context — into a structured career bank, then scores roles and drafts resumes and cover letters tailored to each one.
How it works -> Intelligent LearningLittle Monsters
A K-12 study app — tutor, flashcards, recording, and rewards — designed around a student with dyslexia who built the first version for herself. A real app for a real user.
Read the story -> The rest of the fleet38 applications, one runtime
Operations, World Intelligence, Presentation, Travel, and more — every app loads from a manifest, not a fork, and shares the same runtime, connectors, identity, isolation, and cost control.
Browse the fleet ->Open to senior SAP enterprise architecture, engineering leadership, technical program, and agentic AI platform roles.
Twenty-plus years of enterprise SAP delivery, regulated government-cloud operations since 2021, $82M in Premium Engagement revenue driven scope-to-close — and a working platform you can review. Start with the resume; open oshal / Open Swarm when you want the deeper architecture.