Wireless Engineer III — AAA Development & Platform Automation· Charter Communications (Spectrum)
Nov 2024 – Aug 2026 · Previous role
Wireless Engineer III on a national carrier AAA/RADIUS platform supporting subscriber WiFi authentication, roaming, and accounting workflows.
I joined in integration testing and was promoted into development after six months, then expanded into a cross-cutting platform role spanning telecom policy, infrastructure automation, database reliability, release engineering, security, test platforms, lab operations, and technical enablement.
Across 21 months I took on 5 platform areas; the early-August evidence snapshot records 180 completed tickets and 248 merged merge requests across 31 repositories.
Scope owned concurrently
- AAA deployment automation: versioned Ansible collections for deployment, upgrades, policy changes, auditing, and rollback.
- Cassandra platform operations: backup and restore, topology and upgrade planning, UDF performance optimization, and migration analysis.
- AAA policy engineering: behavioural-parity work for a platform rewrite and accounting workflows.
- Security and compliance automation: endpoint controls, audit-event forwarding, native application SSO, and group-based access.
- Developer and lab enablement: VM provisioning, inventory reconciliation, CI/CD, automated testing, a deployed engineering assistant, and evidence tooling.
Infrastructure automation (Ansible)
- Converted runbook-driven platform deployment into 6 versioned Ansible collections with 29 tagged releases in 76 days, covering deployment, database operations, system administration, VM preparation, orchestration, and preflight checks.
- Built the team's first collection release pipeline and test strategy, including 403 automated tests and 96 property-based tests in the earlier suite snapshot, parallel pytest execution, automatic versioning, artifact publication, and shared CI templates. Later deployment safeguards were challenged at 16 injected failure points and with 13 deliberate role mutations, checking preservation of a startable previous release.
- Decomposed a monolithic deployment role into separate install, policy, and helper roles so routine policy changes could ship without replacing platform software.
- Added dry-run audit reporting so operators could inspect a proposed change before authorizing execution.
- Built a systemd-enabled Docker test harness so role changes could be validated locally instead of consuming shared lab capacity.
- Replaced a multi-step manual deployment handoff with a single-command, version-pinned, idempotent deployment supporting roughly one-minute policy-only updates, dry-run impact reports, post-deploy verification, offline bundles, and sub-second symlink rollback.
Database engineering (Cassandra)
- Owned the Cassandra operations programme and shipped 14 tagged releases in roughly four weeks while the work moved through formal QA certification.
- Automated backup and restore end to end with scheduling, retention, verification, capacity checks, retries, and alerting, replacing an inconsistent manual process.
- Removed a data-integrity footgun in backup naming by adding fail-fast validation and then autodiscovery.
- Implemented mutual-TLS support and service-account-safe certificate discovery for production-equivalent environments.
- Rehearsed a high-risk multi-node topology change against an isolated replica before executing the live change successfully.
- Performed encrypted-data migration analysis, load simulation, and operator-workflow simplification, cutting the planned manual input from roughly twelve variables to four.
- Designed a Cassandra 4.1-to-5.0 rolling-upgrade path for the subscriber-authentication clusters.
- Optimized a legacy Cassandra decryption UDF, reducing development-benchmark mean latency from 12.47 ms to 0.70 ms and increasing 16-thread throughput from 635 to approximately 13,100 operations per second; scoped equivalence and round-trip compatibility checks to the tested variants.
- Released a database-automation fix spanning four CQL roles, correcting credential propagation and variable-precedence defects; tested the published package against operator instructions with a CQL stub and repaired five documentation errors while real-cluster recertification remained pending.
- Diagnosed and closed an availability defect in a subscriber-authentication database where crafted input to its encryption and decryption user-defined functions could bring the cluster down.
CI/CD & release engineering
- Root-caused recurring release blockers across build images, registry routing, and artifact authentication, then replaced repository-local workarounds with shared templates. Reconciled three policy-bundle packaging paths and added checks against overwriting vendor runtime files, resolving inconsistent branch and merge-request validation.
- Stood up a self-hosted CI runner for policy validation and built a multi-architecture base image for arm64 and amd64 development workflows.
- Moved collection publishing onto a shared tag-driven release pipeline with dedicated service credentials and manual approval gates for deployment-sensitive actions.
- Brought the team's repositories to a green compliance pipeline and added automated documentation publication behind approval gates.
- Administered a 12+ repository GitLab group, establishing Git Flow, CODEOWNERS, pipeline-gated merges, branch protection, artifact conventions, and reusable templates across the AAA automation portfolio.
- Migrated a development policy configuration to templated values and encrypted secrets while preserving byte-identical output across all 62 rendered files; credential rotation remained a separate follow-up.
Security & compliance
- Automated deployment of endpoint security, endpoint management, software inventory, vulnerability management, and MFA controls across a pre-production fleet.
- Delivered SIEM log-forwarding automation covering platform and operating-system logs across multiple fleet layouts. Automated enriched Linux audit-event forwarding across six development nodes, resolving SELinux and audit-daemon integration failures and using queued TCP delivery for centralized investigation.
- Remediated tracked vulnerability findings to closure, introduced safer package-manager-native maintenance operations, and delivered a certified upgrade path for legacy Linux hosts.
- Moved source-control permissions onto directory-backed groups and moved CI pipelines off personal credentials onto dedicated service accounts.
- Added repository safeguards that keep lab secrets out of policy source control.
- Automated directory-backed SSO across engineering tools using SAML, OIDC, and OAuth, including group-driven account provisioning and role mapping, tested allow/deny behavior, and preserved local recovery access.
- Shipped CIS-aligned kernel and SSH daemon hardening as a fleet role, validating each generated daemon configuration with a syntax check before applying it and ordering the drop-in files so the existing access policy stayed authoritative instead of being silently overridden.
- Removed a destructive failure mode from fleet kernel maintenance by replacing two fragile shell steps with an explicit running-kernel check and an old-installs-only removal, so the automation could no longer delete the kernel the host had actually booted.
Network & protocol engineering (RADIUS / EAP)
- Reverse-engineered undocumented RADIUS accounting field derivations from policy logs and packet captures, establishing a behavioural-parity baseline for a platform rewrite.
- Rebuilt individual accounting derivations to verified parity against that baseline and created replay tooling for repeatable protocol validation.
- Designed and delivered a state-persistence layer for a telecom policy runtime, then removed 16,390 lines of legacy code using execution evidence to prove what was genuinely unused.
- Diagnosed difficult mutual-TLS and protocol-integration failures across RADIUS-over-TLS and a legacy subscriber-state interface.
- Sole-engineered a ground-up accounting-policy rewrite in the vendor policy DSL: 181 commits in roughly two and a half months across 15+ modules and 30+ configuration files, with CI/CD, defensive protocol parsing, LDAP failover, RADIUS/RadSec forwarding, and a documented 87-code operational logging registry.
- Implemented a single-probe write/read/delete smoke harness for every state model, turning multi-minute manual verification into sub-second post-deploy validation in a DSL with no debugger or stack traces.
- Diagnosed two independent causes of silent RADIUS accounting health-check failures—a runtime compatibility issue and missing protocol attributes—and validated the corrected probe with 12 of 12 successful live responses at approximately 140 ms median latency; load-balancer monitor deployment remained pending.
- Root-caused a TLS handshake failure blocking OAuth2 token retrieval by isolating the ALPN extension in the ClientHello as the trigger, eliminating DNS resolution, TCP reachability, certificate interception, TLS version negotiation and general HTTPS egress with a positive control for each, then proving with a four-variant differential matrix that the reset followed the extension being present rather than the protocol offered, and restoring token issuance the same day.
Test automation & QA
- Designed and shipped a Spring-based web platform for 802.1X/EAP supplicant test automation with configuration generation, credential handling, containerized execution, result parsing, dashboards, SIEM query integration, and automated delivery.
- Architected the platform as eleven Java 23 / Spring Boot modules with nine reusable Maven modules, multi-architecture Docker delivery, and 70 test classes containing more test code than production code, all sole-authored and able to survive 30-way parallel execution.
- Built an end-to-end AAA testing platform spanning a typed Python library, operator CLIs, Android and iOS Appium automation, Docker, and gRPC; it collapsed repeated manual SIEM queries into a single call across 120+ tests and saved an estimated 6–10 engineer-hours in one certification cycle.
- Automated 11 previously manual integration tests, reclaiming about 3.7 engineer-hours per full pass, and built a containerized pre-deployment policy validation gate across seven repositories, where policy bundles must boot in the AAA runtime and clear a 13-pattern fatal-error screen before publication.
- Defined and executed a roughly 30-case release regression suite spanning multiple EAP methods, mobile platforms, roaming scenarios, Passpoint, and negative cases.
- Designed the shared cross-language contract layer (four gRPC services, eight RPC methods, 13 messages, and eight enums) consumed by Java and Python tools for session control, provisioning-cache operations, SIEM queries, and supplicant configuration.
- Built the companion portable task-executor agent in Python spanning four CPU architectures, with Kafka result signals, RADIUS/EAPOL probes, REST task discovery, and CI packaging, targeting a roughly 79-host lab fleet across two data centers and four environment tiers.
- Produced the certification evidence package for a carrier AAA platform release, assembling roughly 90 attachments of policy logs, packet captures, encryption and decryption traces, protocol replay files and device captures across three months.
- Shipped PowerShell automation for reconfiguring wireless settings on Android test devices three weeks into the engagement, extracting profile selection, device serial handling and address validation into a reusable module across nine merged merge requests.
Virtualization & lab platform
- Automated VM provisioning from template preparation through cloud-init self-deployment, including dual-stack IPv4/IPv6 addressing and an IPv6-only defect that had been breaking unattended provisioning. Built and verified a Rocky Linux golden template, then provisioned eight database VMs and checked unique machine identities and SSH host keys on every running guest.
- Built control-node preflight checks that report automation readiness before a run touches infrastructure.
- Delivered supporting engineering services including reverse proxying, status reporting, monitoring alerts, and team alert integrations.
- Validated a major ESXi upgrade on live infrastructure and surfaced a lifecycle risk early enough for it to be resolved before an outage.
- Automated vSphere snapshot restore and environment-aware artifact retrieval, including a SOAP-to-pyVmomi migration, SOCKS5 support, remote DNS, and lab/production artifact-repository fallback so workflows behaved consistently across network segments.
- Built repeatable IPAM import and drift-audit tooling, reconciled a 233-VM development inventory, and populated a separate integration-test inventory covering 62 VLANs, 144 subnets, and 176 addresses using hypervisor, host, and telemetry evidence.
- Restored an internal application's login service by tracing failed database session writes to disk exhaustion caused by a cascading Redis and container-logging failure; reclaimed capacity and prepared log-rotation controls.
Documentation & enablement
- Authored 69 internal wiki pages over the engagement, including installation guides carried through formal QA certification.
- Wrote an 11-page developer-onboarding suite for an internal policy language and runtime covering language reference, conventions, repository structure, testing, secrets handling, releases, and troubleshooting.
- Wrote internal proposals that set direction for collection-based automation, automated deployment, and software lifecycle planning.
- Operationalized a 935-note engineering knowledge base as team documentation, publishing a curated 229-note subset behind an explicit per-note publication gate.
- Mentored engineers on Git and tooling and built an IDE plugin for the platform's policy language.
- Built and deployed a TypeScript Webex assistant for engineering knowledge lookup, with access configured for 12 teammates, separate read-only tool gateways, and conversation isolation by caller and privilege tier; verified lookup and refusal behavior through live messaging.
- Built reusable AI-workflow tools for engineering evidence retrieval, outbound sanitization, and document consistency, with 193 passing assertions and negative tests; corrected repository-counting errors caused by duplicate counting of repository history and uninitialized submodules.
- Built and maintained a library of roughly 54 reusable AI agent skills across three engineering workspaces, codifying workflows for infrastructure operations, investigation and evidence gathering, security remediation, documentation, configuration management and telecom policy work, and authored an audit that reconciles all three workspace roots in a single run to flag duplicated skills, dead bundled-file references, one-sided skill boundaries, and colliding trigger phrases that would otherwise make agent dispatch nondeterministic.
- Carried the platform automation portfolio as its sole maintainer in 21 of the 27 codebases surveyed, spanning configuration management, telecom policy, test platforms and operational tooling with no second commit author.
Infrastructure automation (Ansible) (in progress)
- Designed deployment-time configuration and secrets tooling that separates reusable policy artifacts from environment values and versioned encrypted secrets; released the deployment component with environment validation, provenance tracking, and compatibility checks while broader rollout remained in progress.
Documentation & enablement (prototype)
- Prototyped a per-engineer status ledger using two weeks of real work, identified conflicts in hand-maintained reporting, and designed a refreshable contribution model around existing Jira fields; recurring automation and broader adoption remained next steps.
Database engineering (Cassandra) (in progress)
- Piloted a Kubernetes operator deployment model for a carrier subscriber-authentication database tier and released a dedicated collection for it, with the pilot still open at the end of the evidence window.
Security & compliance (in progress)
- Scoped an environment-wide security remediation programme into seven tracked work items covering agent staging, operating-system patching, MFA, SIEM forwarding and a follow-up security assessment, delivered the first of them, and kept the rest moving while one item stayed blocked on an approval process outside his control.
- Ansible
- Cassandra
- Nokia AAA / TAL
- RADIUS / RadSec
- EAP-AKA / EAP-TTLS
- 802.1X
- Passpoint
- GitLab CI
- Artifactory
- Docker
- Kubernetes
- Splunk
- VMware vSphere
- cloud-init
- RHEL / Rocky
- Ubuntu
- Python
- Java 23 / Spring Boot
- Appium
- gRPC / proto3
- PowerShell
- TypeScript
- Webex
- MCP
- Authentik
- SAML / OIDC
- phpIPAM
- SOPS / age