Robert Myhre
Solutions Architect
Python
Experimentation & Evaluation: 5/10
Active 1 day ago
Invite to interview
Download CVCV
Overview
Technical skills
Projects
Timeline
Roles
Overview
Platform-focused backend engineer (senior) specializing in reliable, test-driven AI operational tooling and governed orchestration. Their strongest proven skill is designing deterministic, auditable orchestration and recovery semantics as shown by the exhaustive tests that pin the acquisition-floor and recovery-decision invariants (project-boundary-evidence/validation/test_acquisition_floor.py and test_recovery_decision.py). The public code has limited evidence of production-scale service orchestration patterns like distributed rate limiting, schema migration histories, or multi-service contract definitions that would prove cross-service system-design leadership.
Phone
Technical skills
Languages
1
Python
DevOps
4
AWS
Azure
Kubernetes
GitHub
Other
16
Zero Trust
Pydantic
Git
Terraform
Rest API
AI Agents
Ansible
AI/ML
Frontend
RAG
LLM Guardrails
Multi-Agent Systems
Knowledge Graph
BGP
MPLS
OSPF
Projects
Enterprise AI Infrastructure Architecture
Conducted architecture-focused research on GPU compute and high-performance networking for distributed AI systems. Built multi-GPU environments to explore workload placement and performance factors such as NUMA locality, GPU/NIC affinity, and congestion behavior. Translated infrastructure and operational constraints into repeatable infrastructure patterns with telemetry and observability considerations.
Reasoning Context Admission — Architecture Building Block
Developed a reusable architecture pattern that governs what information AI systems can treat as trusted reasoning context. Defined guidance for semantic screening, blast-radius decision constraints, and lifecycle governance for agentic applications and AI-enabled automation. Extended the pattern to support RAG-style systems and autonomous workflow reasoning while maintaining governance controls.
Boundary — Governed Agentic Infrastructure Reference Architecture
Designed and implemented a working reference-architecture prototype for safely integrating autonomous AI agents with enterprise infrastructure. Created reusable principles and decision guidance around trust, admission control, blast-radius boundaries, and deterministic guardrails, separating probabilistic reasoning from deterministic control. Built multi-agent orchestration with policy enforcement, identity, observability, and controlled infrastructure mutation, supported by automated testing and conformance checks.
Timeline
IT Principal Architect
•
Lead
Banner Health
•
Full-Time
Serves as a voting member of the Solutions Architecture Council, reviewing logical architectures and identifying cross-domain risks and gaps. Develops and governs enterprise architecture standards, including cloud IP addressing and zone-segmentation patterns deployed across many sites. Partners with enterprise and security architecture teams to create reference architectures and decision guidance, leads technology evaluations and PoC/PoV efforts, and designs multi-cloud Network-as-a-Service connectivity for AWS/Azure with redundant regional paths.
AWS
Azure
Zero Trust
Senior Solutions Architect
•
Lead
Evolving Solutions
•
Full-Time
Provides customer-facing architecture support for enterprise data-center, cloud, networking, and automation offerings from discovery through proposals and delivery. Translates business and organizational constraints into multiple architecture alternatives, including data-center design tradeoffs and migration planning. Architected resilient data-center and cloud solutions covering security segmentation and network designs using AWS/Azure networking, taking architectures through implementation and operational transition.
AWS
Azure
Principal Network Architect
•
Lead
Sleep Number
•
Full-Time
Evolved from network architecture into the company’s cross-domain architecture organization. Led the architecture and migration strategy to transform a locally hosted data center into two geographically redundant colocation facilities, defining target network resiliency and connectivity patterns. Facilitated cross-functional workshops to assess dependencies and select network technologies, and carried the resulting connectivity and carrier-path resiliency architecture through implementation and migration planning.
Architecture (Data Center Modernization)
•
Lead
Red Wing Shoes
•
Full-Time
Led on-premises data-center modernization and helped assess whether increased complexity was justified for an architecture change. Recommended a traditional Cisco Nexus approach over ACI based on business and operational requirements, then supported implementation of the selected design. Advised technical and CIO-level stakeholders on architecture decisions and execution.
Middle AI/ML Engineer
Confidence: Medium Generalist
Governance-oriented systems engineer and test-driven architect working at a senior-leaning level with deep attention to deterministic orchestrator invariants and auditability. The strongest proven skill is designing and validating deterministic governance primitives and durable artifact behaviors as shown by the exhaustive tests in validation/test_acquisition_floor.py and validation/test_recovery_decision.py. Public code does not contain model-training loops, large-scale data pipelines, or production serving infrastructure, so GPU optimization, custom training, and deployment pipelines are not evidenced here.
Model Architecture & Training
1/10
How well models are designed and trained
Minimal direct model-building or training artifacts in the human-authored files; tests assert orchestration boundaries and ensure no model calls are made by the deterministic gates.
Evidence
project-boundary-evidence/validation/test_acquisition_floor.py: uses run_floor, GatewayBudget and asserts ordering of SOT_HOST_BY_NAME_TOOL and LIVE_FACTS_TOOL
project-boundary-evidence/validation/test_recovery_decision.py: composes recovery decision artifacts and validates disposition constraints
Data Pipeline & Feature Engineering
2/10
How data is prepared for models
Light data-handling evidence focused on deterministic acquisition and source provenance rather than feature engineering or large-scale ETL.
Evidence
project-boundary-evidence/validation/test_acquisition_floor.py: sources(tmp_path) fixture writing live-facts YAML and corpus index
project-boundary-evidence/validation/test_acquisition_floor.py: derive_targets contract scanning rules for hostname extraction
Experimentation & Evaluation
5/10
How results are measured and tested
Strong, well-structured experimentation and evaluation evidence in the form of comprehensive, architecture-level tests that validate invariants, failure modes, and idempotency; clear test coverage for key governance behaviors.
Evidence
project-boundary-evidence/validation/test_acquisition_floor.py: multiple parametrized tests covering absence semantics, budget halting, and idempotency of floor invocations
project-boundary-evidence/validation/test_recovery_decision.py: tests for composing, persisting, loading, and conflict detection of recovery decision artifacts
MLOps & Deployment
3/10
How models are shipped to production
Orchestration and operational artifacts around gated acquisition and artifact persistence indicate MLOps thinking, but deployment and monitoring pipelines are not present in the human-authored files.
Evidence
project-boundary-evidence/validation/test_acquisition_floor.py: asserts stage_accounts, StageAccount construction, and gateway error propagation
project-boundary-evidence/validation/test_recovery_decision.py: artifact path composition and persist/load semantics
Computational Efficiency
How efficiently computing resources are used
Not evidenced in public code
Research Depth & Innovation
2/10
Depth of research and new ideas
Design demonstrates attention to architecture invariants, provenance, and safety-minded rules but does not implement novel algorithmic research or reproduce papers in the human-authored artifacts.
Evidence
project-boundary-evidence/validation/test_acquisition_floor.py: architecture-invariant tests (acquisition ordering, lane labeling)
project-boundary-evidence/validation/test_recovery_decision.py: enforcement of per-act recovery artifact identity and refusal semantics
Expertise
AI Agents & Agentic Workflows• Middle
Industries
Artificial Intelligence• Middle
Software• Middle
Technologies
AI/ML
Terraform
Ansible
AI Agents
Azure• since 2021
Git
AWS• since 2021
Kubernetes
RAG
GitHub
LLM Guardrails
Multi-Agent Systems
BGP
OSPF
MPLS
Human-in-the-Loop• mentioned only
Llama• mentioned only
LLM• mentioned only
Mixture of Experts• mentioned only
Observability• mentioned only
Systems• mentioned only
vLLM• mentioned only
Recommendations
- Develop governed agentic workflows and orchestrator modules where deterministic acquisition, provenance tagging, and per-act authorization matter - build the orchestrator and test harnesses for each gate
- Design and implement recovery and audit artifact persistence layers that enforce idempotency and tamper-evidence - expand the recovery_decision persistence interface into a small service with end-to-end integration tests
- Implement reproducible lab-run instrumentation and evaluation dashboards for the acquisition floor tests - capture run-level provenance, structured metrics, and automated regression checks
- Own evidence-driven safety features around gate budgets and acquisition defects - translate tests into CI gates that prevent regressions in the deterministic acquisition semantics
Repositories
The developer's experience in this domain has been verified based on AI analysis of the following repositories:
Middle Backend Developer
Confidence: Medium Platform Engineer
Platform-focused backend engineer (senior) specializing in reliable, test-driven AI operational tooling and governed orchestration. Their strongest proven skill is designing deterministic, auditable orchestration and recovery semantics as shown by the exhaustive tests that pin the acquisition-floor and recovery-decision invariants (project-boundary-evidence/validation/test_acquisition_floor.py and test_recovery_decision.py). The public code has limited evidence of production-scale service orchestration patterns like distributed rate limiting, schema migration histories, or multi-service contract definitions that would prove cross-service system-design leadership.
API Design
4/10
How well APIs are designed
API design shows deliberate, consistent contracts around deterministic prompts and labeled acquisition records but lacks an explicit public versioning strategy or documented idempotency keys on external REST boundaries.
Evidence
project-boundary-evidence/validation/test_acquisition_floor.py: asserts call ordering, acquisition labels and prompt_templates rendering (context_scope.SCOPE_TEMPLATE)
project-boundary-evidence/validation/test_acquisition_floor.py: tests that floor evidence is placed in the first data block and labeled with acquisition='floor'
Data Layer & Database
3/10
Working with databases
Data-layer work emphasizes durable artifact naming, idempotent persistence, and explicit absent-record handling rather than complex DB schema evolution or tuned SQL.
Evidence
project-boundary-evidence/validation/test_recovery_decision.py: tests persistence idempotency and conflict detection around recovery-decision artifacts (compose_recovery_decision, persist)
project-boundary-evidence/validation/test_acquisition_floor.py: uses filesystem-backed SourceContext with ledger, live-facts YAML files and asserts recorded absent-targets
Scalability & Performance
2/10
Handling load and speed
Some operational scaling and budgeting controls are present (gateway budgets and pre-flight halting) but there is no public evidence of system-level caching/invalidation, rate limiting, or measured load-optimizations in the human-authored files.
Evidence
project-boundary-evidence/validation/test_acquisition_floor.py: GatewayBudget tests (floor budget insufficient halts before model calls)
project-boundary-evidence/validation/test_acquisition_floor.py: tests that budget accounting separates floor and agent lanes
System Architecture
5/10
Overall system structure
Clear architectural decisions are codified with tests that pin invariants (deterministic acquisition floor, separation of authority, recovery-as-separate-artifact) showing design thinking about boundaries and failure modes.
Evidence
project-boundary-evidence/decisions/README.md: enumerates architecture-change records and the deterministic acquisition-floor requirement
project-boundary-evidence/validation/test_acquisition_floor.py: end-to-end tests that pin the acquisition-floor invariants
Security & Auth
5/10
Protecting data and access
Strong security and authorization attention is visible: separation of source-of-truth vs live facts, explicit failure classes, and recovery authorization mechanics are enforced and tested.
Evidence
project-boundary-evidence/validation/test_acquisition_floor.py: tests that a Source-of-Truth absence never suppresses live-facts and that ambiguous SoT matches are treated as defects
project-boundary-evidence/validation/test_recovery_decision.py: tests that recovery decisions are separate artifacts and that policy-authorized runs can still compose recovery authorizations
Reliability & Observability
6/10
Stability and monitoring
Reliability and observability are strong priorities: extensive tests assert idempotency, durable stage accounts, explicit error propagation, and re-run behavior; proofs in tests show invariants are observable and auditable.
Evidence
project-boundary-evidence/validation/test_recovery_decision.py: tests for write-once behavior, conflict detection, and read-back-first persistence
project-boundary-evidence/validation/test_acquisition_floor.py: tests that stage accounts are produced and that halted runs preserve call records; idempotency tests for re-run after a halted floor
Expertise
Backend AI & LLM• Middle
Python• Middle
Industries
Artificial Intelligence• Middle
Technologies
Python• Middle
Rest API
Pydantic
Recommendations
- Build governed AI-operation tooling that requires deterministic inputs and auditable outputs - e.g., orchestrators enforcing Source-of-Truth vs live-facts policies.
- Lead development of safety-critical, test-first backend components such as deterministic gating, artifact composition, and write-once persistence for authorization flows.
- Develop observability and reproducibility tooling for ML infra - launch/bringup scripts, NVML sampling, and evidence-capture frameworks are a strong match.
- Work on contract-driven backend systems where idempotency, explicit failure classifications, and durable stage accounts are required.
Repositories
The developer's experience in this domain has been verified based on AI analysis of the following repositories:
Middle Frontend Developer
Confidence: Medium UI Engineer
A frontend-focused UI engineer (approx. middle level) who specializes in well-crafted documentation theming and CSS-driven presentation. The strongest proven skill is building polished, responsive documentation and design shells as evidenced by docs/assets/styles/extra.css and the print.css rules. There is little or no evidence of component-level JavaScript, client-state management, tests, or runtime performance instrumentation in the human-authored code.
UI Component Architecture
2/10
How interface parts are built
Styling is organized and scoped into reusable visual blocks but there is no component code or runtime composition logic; architecture is visual/CSS-first rather than component-driven.
Evidence
docs/assets/styles/extra.css: many scoped class groups resembling UI components (snapshot-hero, snapshot-card, boundary-question-card, boundary-architecture__node)
overrides/main.html: Jinja template extending base.html and providing site-level HTML composition and meta blocks
Responsive & Cross-browser
6/10
Works on all screens and browsers
Clear responsive work with clamp(), multiple media queries, print-specific styles, and prefers-color-scheme handling; shows deliberate layout and breakpoints for different viewports.
Evidence
docs/assets/styles/extra.css: use of clamp() on .snapshot-name and .md-typeset > h1, multiple @media blocks for breakpoints and layout adjustments
docs/assets/styles/print.css: print-oriented media rules and @page size/margins for A4
Performance Optimization
2/10
Speed of the interface
No measurable performance instrumentation or JS-driven optimizations are present; CSS shows pragmatic choices that avoid heavy runtime effects and includes print compaction rules.
Evidence
docs/assets/styles/extra.css: restrained use of box-shadow and transitions with reduced-motion awareness, and print-specific simplifications
docs/assets/styles/print.css: explicit print optimizations that remove heavy layout chrome and shadows
Accessibility & Semantics
4/10
Usable for everyone
Some accessibility-minded touches in styling and focus handling are present, but no systematic a11y enforcement or ARIA-enabled custom widgets in the human-authored files.
Evidence
docs/assets/styles/extra.css: .boundary-evidence-table:focus rule giving an outline for keyboard focus
docs/assets/styles/extra.css: targeted rules for PDF links and image figure captions improving semantics and readability (.md-typeset > p > a[href$=".pdf"] and figure figcaption styles)
State Management & Data Flow
1/10
Managing data in the app
No client-side state management, async discipline, request cancellation, or state machines are present; site-level template is static HTML/Jinja without data-flow logic.
Evidence
overrides/main.html: Jinja template that controls meta/head composition, with no client-side state or async logic present in the human-authored templates
UX & Visual Polish
6/10
Look and feel quality
High visual polish and UX attention in typography, spacing, hero treatment, and micro-interactions; consistent patterns for loading/print/responsive states in layout and document density.
Evidence
docs/assets/styles/extra.css: snapshot-hero, snapshot-actions and snapshot-card rules providing polished hero, actions and hover transitions
docs/assets/styles/extra.css: document shell rules (md-typeset > h1/h2/h3, blockquote, tables) that improve long-form reading and presentation
Expertise
HTML & CSS• Middle
Industries
Artificial Intelligence• Middle
Software• Middle
Technologies
Frontend
Cloud• mentioned only
Systems• mentioned only
Recommendations
- Design and implement documentation theming and MkDocs/templating work where meticulous CSS, print and responsive behavior are required.
- Create and maintain a CSS-based design system or scope tokens and component styles for documentation and marketing websites.
- Polish landing pages, typographic systems, and accessible print styles for technical content and long-form architecture docs.
Repositories
The developer's experience in this domain has been verified based on AI analysis of the following repositories: