# Clark Ohlenbusch — Full Technical Dossier & Systems Specification

> Author: Clark Ohlenbusch  
> Canonical URL: https://clarkohlenbusch.com/llms-full.txt  
> Updated: 2026-09-03  
> Contact: clark.ohlenbusch@gmail.com  
> Location: Boston, Massachusetts, USA  

---

## 1. Professional Overview & Technical Thesis

Clark Ohlenbusch is a software engineer, product manager, and systems security researcher operating at the intersection of three rigorous computing disciplines:
1. **Agentic AI Security & Runtime Governance**: Designing deterministic runtime authorization, zero-trust boundary controls, and cryptographic delegation chains for autonomous multi-agent systems.
2. **Real-Time Multimodal Voice AI**: Building full-duplex, low-latency audio streaming engines, conversational prompt graph editors, and dialectic dialogue systems for high-stakes enterprise, education, and clinical applications.
3. **Systems Programming & Formal Verification**: Low-level memory architecture, pointer safety, POSIX multithreading, and proof verification using interactive theorem provers (Lean 4 and Coq).

The unifying philosophy of Clark's work is **high-integrity systems**: ensuring that as software increases in autonomy and multimodal capability, its boundaries remain formally verifiable, deterministic, and safe.

---

## 2. Detailed Work History & Technical Specifications

### 2.1 Saviynt Inc. (August 2023 – Present)
- **Role**: Product Manager & Forward Deployed Engineer
- **Domain**: Agentic AI Security (Zuma Platform)
- **Location**: Boston, MA / Remote
- **Technical Specifications & Deliverables**:
  - **Zuma Platform Architecture**: Product and engineering leadership from inception through General Availability (GA). Authored PRFAQs for builder experiences (agent SDKs, CLI, secret scanning, and automated dev-to-prod promotion workflows).
  - **Runtime Authorization Engine**: Engineered policy enforcement middleware using Open Policy Agent (OPA) with Rego policies. The engine dynamically evaluates intent deviation, validates pre-tool and post-tool webhook payloads, and intercepts anomalous agent behavior before external API execution.
  - **Segregation of Duties (SOD) for Agent Chains**: Designed design-time and runtime SOD matrix enforcement across multi-agent workflows (A2A and H2A). Prevented toxic permission aggregations (e.g. an autonomous agent reading financial transaction records while possessing write permissions to payment dispatch APIs).
  - **Multi-Cloud Discovery & Gateway Orchestration**: Defined specifications for discovering and cataloging autonomous agents across AWS Bedrock, Google Cloud Vertex AI, Microsoft Azure AI Foundry, Microsoft Copilot Studio, and Salesforce Agentforce.
  - **Enterprise Commercial Impact**: Conducted live technical proofs-of-concept (POCs) for major global financial institutions (Northern Trust, FIS). Authored technical architecture sections for 22+ enterprise RFPs and RFIs, securing ~$15M in closed-won deal value.
  - **Workforce Enablement**: Created and delivered the 'Agentic50' technical training curriculum to 100+ global systems engineers covering Model Context Protocol (MCP) deployment, Okta OAuth token exchange, and Copilot Studio integrations.

### 2.2 Spoqen (September 2025 – March 2026)
- **Role**: Co-Founder & Product Lead
- **Domain**: Voice AI Autonomous Reservation Agents & Multi-Agent Workflow Compilers
- **Location**: Boston, MA / Remote
- **Technical Specifications & Deliverables**:
  - **Origin & Leadership**: Spun company out of placing 2nd in North America (out of 80+ teams) at the AstraZeneca Global Health Hackathon with an acoustic voice AI agent. Led a distributed 5–6 person engineering team across sprint execution, product discovery, and production architecture.
  - **Autonomous Booking Agent (ResNexus CRM)**: Deployed 24/7 autonomous voice reservation agent integrated directly into ResNexus CRM for RV resort operators. Masked 95% of the CRM automation surface to guarantee a 100% call-resolution metric (65–75% autonomous completion + zero-latency human handoff).
  - **Multi-Agent Transcript-to-Flow Compiler**: Architected multi-agent graph compiler (LangSmith + Gemini 2.5 Flash 1M context) converting raw client call recordings into visual DAG canvases, discovering unmapped edge-case call paths and slashing voice workflow authoring from days to minutes.
  - **Voice Stack & Telemetry Optimization**: Evaluated production voice infrastructure across VAPI vs. Retell and Cartesia Sonic; optimized Claude 3.5 Haiku system prompts to ~15K tokens for sub-second tool execution and migrated event pipelines to Pusher WebSockets with Sentry telemetry to reduce compute costs.
  - **Commercial Partnerships & Unit Economics**: Negotiated strategic partnership with Upside AI accessing 13,000 RV park operators, anchoring $4,500/month pricing against $50K/year human receptionist payroll economics; advanced enterprise pilot discovery with Stack (~500 residential units).
  - **Venture Conclusion & IP Asset Sale**: Concluded venture in March 2026 following a successful proprietary IP asset sale, transferring core voice agent automation codebase, multi-agent compiler architecture, and booking pipeline to an enterprise acquirer.

### 2.3 Meno (LearnMeno) (2025 – 2026)
- **Role**: Creator & Lead Architect (Co-Founder)
- **Domain**: Socratic Voice AI Dialectic Learning & Institutional Assessment
- **Location**: Boston, MA / Remote (EdTech Venture)
- **Technical Specifications & Deliverables**:
  - **Founder Thesis & Product Execution**: Founded and architected LearnMeno ("A more human way to learn", inspired by Plato's *Meno*) to reimagine higher-education learning and assessment for the modern era through real-time voice AI and Socratic oral examination, where students verbally articulate and deepen conceptual mastery.
  - **University Institutional Pilots**: Secured active university pilots at Boston College and UMass Boston (with fall pilot commitment at Colorado Mesa University); navigated university IT security clearances, FERPA compliance, and student AI-consent pathways for lecture cohorts up to 250 students.
  - **Full-Duplex Socratic Voice Pipeline**: Engineered real-time audio-to-audio dialectic assessment engine on Google Vertex AI; benchmarked Whisper Large v3 Turbo (faster-whisper), Kokoro TTS, and Gemini Live API to deliver sub-second turn-taking, natural interruption (barge-in) handling, and verbal comprehension scoring.
  - **Zero-Friction LMS & Educator Platform**: Architected Canvas and Blackboard roster integration via Google OAuth, enabling students to launch oral assessments via dynamic links with zero account creation friction; deployed professor dashboards with tri-mode AI rubric grading (Manual, AI-Review, Instant AI).
  - **Customer Discovery & Capital Strategy**: Conducted 30+ professor interviews identifying oral exam grading as the #1 institutional pain point, pivoting GTM from individual instructors to department chairs and Centers for Teaching & Learning; pitched SSC Venture Partners ($10M fund).

### 2.4 Northeastern University — CHAI Lab (October 2025 – Present)
- **Role**: Voice AI & HCI Researcher
- **Domain**: Conversational Formality, Acoustic Pacing & Older Adult HCI
- **Location**: Boston, MA
- **Technical Specifications & Deliverables**:
  - **Empirical Clinical Trial (n=140)**: Deployed a large-scale human-computer interaction trial using Prolific. Investigated how conversational formality levels, acoustic pause duration, and turn-taking latency impact trust, comprehension, and cognitive ease in older adult demographics.
  - **Real-Time Telemetry Dashboard**: Engineered a full-stack telemetry and analytics dashboard tracking p-values, formality classification metrics, audio waveform playback, and transcript session replays for longitudinal evaluation.
  - **Conference Publication**: Synthesized trial methodologies and empirical results for submission as an ACM CHI / CUI (Conversational User Interfaces) conference poster.

### 2.5 Calico Care (October 2024 – Present)
- **Role**: AI Development Consultant
- **Domain**: Clinical Voice Outreach & Healthcare Automation
- **Location**: Boston, MA
- **Technical Specifications & Deliverables**:
  - **HIPAA-Compliant Voice Pipeline**: Architected HIPAA-compliant telephony pipelines using VAPI and AWS S3 encrypted audio storage, replacing legacy Amazon Alexa-based patient outreach workflows for COPD and CHF cohorts.
  - **Automated Clinical Testing Suite**: Built a 12+ scenario automated evaluation pipeline validating clinical response guardrails, emergency symptom detection, and conversational safety.
  - **Multilingual Support**: Deployed multilingual voice agents in Portuguese and German, evaluating Electronic Medical Record (EMR) integration touchpoints.

### 2.6 Town of Weymouth (May 2025 – August 2025)
- **Role**: Security Engineering Consultant (Freelance)
- **Domain**: Enterprise Authentication & Identity Security
- **Location**: Weymouth, MA
- **Technical Specifications & Deliverables**:
  - **Enterprise YubiKey MFA Rollout**: Designed and deployed hardware-backed FIDO2/WebAuthn YubiKey multi-factor authentication across a municipal Windows enterprise environment (Active Directory and ADFS).
  - **Operational Runbooks**: Authored technical runbooks, disaster recovery procedures, and administrative enablement documentation for municipal IT personnel.

### 2.7 University of Massachusetts Boston (January 2024 – February 2026)
- **Role**: Teaching Assistant (CS240: Coding in C) & Undergraduate Researcher
- **Location**: Boston, MA
- **Technical Specifications & Deliverables**:
  - **Instructional Mentorship**: Led laboratory sections and office hours across 5 academic semesters. Mentored 100+ computer science students in manual heap memory allocation (`malloc`, `calloc`, `realloc`, `free`), pointer arithmetic, double pointers, stack frames, segmentation fault triage (via GDB and Valgrind), and POSIX file descriptors.
  - **Delegation Provenance Records (DPR)**: Co-authored research establishing DPR, a zero-trust governance protocol providing cryptographic lineage verification and dynamic scope limitation for autonomous multi-agent execution graphs.
  - **Theorem Automation (Lean 4 & Coq)**: Researched foundation model tactical proof generation, deterministically validating synthesized proofs through the Lean 4 and Coq kernel checkers with zero hallucination.
  - **Static Analysis & Data Races (NEPLS at BU)**: Evaluated LLMs' ability to detect and patch POSIX multithreading data races using static analysis tools orchestrated over the Model Context Protocol (MCP); presented at the New England Programming Languages and Systems Symposium at Boston University.

---

## 3. Academic Credentials & Verified Honors

- **Institution**: University of Massachusetts Boston
- **Degree Conferred**: Bachelor of Arts in Computer Science
- **Cumulative GPA**: 3.58 / 4.00
- **Honors Ledger**:
  - **John F. Kennedy (JFK) Award Nominee**: University-wide award honoring students who embody civic engagement, intellectual achievement, and leadership.
  - **Dean's List**: 5 Consecutive Semesters of Academic Distinction.
  - **Elizabeth Hanopol Merit Scholarship**: Competitive merit scholarship awarded for computing excellence.
  - **Student Senate Merit Scholarship**: Awarded for instructional leadership and mentorship as CS240 Teaching Assistant.
  - **AstraZeneca Global Health Hackathon**: Runner-Up (2nd Place Global) for acoustic vocal biomarker agent detecting early-onset Alzheimer's.
  - **UMass Boston AI Symposium Hackathon**: First Place & Crowd Favorite ($1,000 award) for respiratory distress acoustic classifier.

---

## 4. Technical Skills & Systems Matrix

- **Programming Languages**: C (Low-Level Systems, POSIX, Memory Management), Rust, TypeScript, JavaScript, Python, Bash/Shell, SQL, Rego (OPA), Lean 4, Coq.
- **AI Security & Governance**: Agentic AI Security, Open Policy Agent (OPA / Rego), Model Context Protocol (MCP), Delegation Provenance Records (DPR), Multi-Agent Segregation of Duties (SOD), JWT-Signed Webhooks, OAuth 2.0 (Okta, Salesforce), Zero-Trust Architecture.
- **Real-Time Voice & Audio Systems**: VAPI, WebSockets, Full-Duplex Audio, Acoustic Turn-Taking, Telephony (Twilio, SIP), Socratic Dialogue Models, HIPAA Audio Compliance, Visual Workflow Graph Synthesis.
- **Cloud Infrastructure & Data**: AWS (Bedrock, S3, IAM, Lambda), Google Cloud Vertex AI, Microsoft Azure AI Foundry, Docker, Redis, Supabase (PostgreSQL with Row-Level Security), Linux Administration (Debian, Ubuntu), Tableau BI Telemetry.

---

## 5. Contact & Canonical Coordinates

- **Email**: clark.ohlenbusch@gmail.com
- **Website**: https://clarkohlenbusch.com
- **Curriculum Vitae**: https://clarkohlenbusch.com/resume
- **Resume AST (Abstract Syntax Tree)**: https://clarkohlenbusch.com/resume.ast.json
- **JSON Resume (Standard Schema)**: https://clarkohlenbusch.com/resume.json
- **PDF Resume**: https://clarkohlenbusch.com/clark-ohlenbusch-resume.pdf
- **GitHub**: https://github.com/ClarkOhlenbusch
- **LinkedIn**: https://www.linkedin.com/in/clark-ohlenbusch-bb8b60253/