$ cat ops.md | head
OPERATOR_AVAILABLE

AI FOR CONTACT CENTERS,
BUILT BY SOMEONE WHO'S RUN ONE.

// 15 years running an outsourced ISP technical-support BPO at OneSupport — 14 ISP clients, 400+ agents, 21 states. Now building production AI that reduces contact rates, automates notes and disposition, and replaces tier-0 calls.

> ./status --credibility
  • 14
    [ ISP / DSL CLIENTS MANAGED ]
  • 295 → 137
    [ AGENTS / 28-MONTH TRANSFORMATION ]
  • -194s
    [ AHT CUT / COMPANY ALL-TIME RECORD ]
  • $1.7M / YR
    [ TRUCK-ROLL SAVINGS / ONE ISP CLIENT ]
> ./build --target=noc

// WHAT I'D BUILD FOR YOUR NOC

AI-Assisted Notes & Disposition

100% QA coverage. Agents focus on the call, not the keyboard.

  • STT captures every word. LLM generates structured notes + classified disposition + QA scorecard.
  • Agents review and commit with one click. Confidence threshold gates auto-commit.
  • Replaces 2–5% manual QA sampling with 100% rubric-scored coverage.
Deepgram Claude Haiku Forced Tool Use Custom Vocab

Tier-0 Voice Deflection

Reduce contact volume at the IVR layer before a human agent answers.

  • Sub-2s end-to-end latency voice agent with a voice clone.
  • Handles status checks, appointment confirmation, after-hours intake, power-cycle scripts.
  • Graceful handoff to human when escalation needed.
Pipecat Deepgram nova-3 Claude Sonnet ElevenLabs

Real-Time Interaction Scoring

Every call, SMS, email, and chat scored 0–100 in real time.

  • Detects repeat callers, sentiment shifts, escalation risk before they bubble up.
  • Root-cause taxonomy rolls dispositions up to engineering tickets — closed-loop RCA.
  • Under $0.03 per call all-in. Cost-controlled at scale.
Claude Haiku 4.5 Prompt Caching Bounded JSON Structured Outputs
> ./history --employer=onesupport

// ONESUPPORT — 15 YEARS RUNNING AN ISP SUPPORT BPO

> ./narrative

Progressed from frontline DSL technician to Director of Operations at OneSupport, a multi-LOB BPO running outsourced ISP technical support for 14 multi-state clients across 21 states. At peak: 400+ agents, $11.5M annual business-unit revenue, full P&L ownership, direct executive reporting to CEO and VPs. Led a 28-month operational transformation that set the company's all-time records for AHT and FCR while cutting staffing by more than half — proof that I've already done at scale what I'm proposing to do with AI now.

[ ANALOG ]

ISP / DSL / broadband Tier 1+2 technical support. Network-issue diagnosis from call data. The exact analog to the BPO labor an enterprise NOC outsources today.

Reduced staffing 295 → 137 agents (-54%)
Cut AHT by 194 seconds (fastest in the company's 25-year history)
Raised FCR by 8% to 78.3% (highest in the company's 25-year history)
Lifted adherence from 90.3% to 95.7%
$2M+ in annual staffing savings
$1.7M / year saved for one ISP client via truck-roll reduction process
Onboarded 137 Tier I agents in 3 months during COVID surge
Most efficient of 7 business units company-wide, 2018 + 2019
> ./ps --running

// WHAT'S RUNNING TODAY AT MAC SEPTIC

AI Interaction Analyzer

[ LIVE · 90 DAYS UPTIME ]
  • Real-time scoring of every call, SMS, email, chat at Mac Septic.
  • Deepgram nova-3 → Claude Haiku 4.5 + Sonnet 4.6 → bounded JSON.
  • Auto-suppression of unsubscribes, weekly Opus strategist summaries.
  • Under $0.005 per scored interaction with prompt caching.

Voice Agent Phil

[ LIVE · SUB-2S LATENCY ]
  • Pipecat + Deepgram + Claude + ElevenLabs voice clone.
  • Handles inbound: appointments, status checks, after-hours intake.
  • ~60% margin per call. Prototype for tier-0 contact-rate reduction.

Production Cross-Channel CRM

[ LIVE · SINGLE OPERATOR, AI-ORCHESTRATED ]
  • FastAPI + Postgres 16 + React 19 + TanStack Query.
  • 70+ route modules, async SQLAlchemy 2.0, WebSocket real-time, Redis with circuit-breaker fallback.
  • Built by orchestrating AI with human-in-the-loop guardrails.
> ./whoami

// ABOUT WILL

> ./id_card
WB
WILL_BURNS
// OPERATOR + AI ORCHESTRATOR

18 years of operations leadership across BPO, MSP, and field services. Texas-based, San Marcos. Built and rebuilt teams that handle the inbound load nobody else wants to own.

Currently Head of TX Operations at Mac Septic, running multi-state field services AND building the AI stack the company runs on — cross-channel CRM, real-time interaction scoring, voice agent handling inbound calls.

AI orchestrator: directs production AI tooling (Claude, Deepgram, Pipecat, custom orchestrators) to ship operational software. Not a from-scratch engineer — an operator who uses AI to build the tools my work requires.

// BJJ blue belt · Texas State Rugby Hall of Fame · U13 basketball coach.

[ LOCATION ]
San Marcos, TX
[ CURRENT_ROLE ]
Head of TX Ops, Mac Septic
[ TENURE ]
18 yrs ops leadership
[ DOMAINS ]
BPO · MSP · Field Services
> ./contact --open

LET'S REDUCE YOUR
CONTACT RATE.

// Available for senior NOC, contact-center, and customer-experience roles — remote or Texas-based. Bring the inbound problem. I'll bring the operators-eye plan and the AI to back it.