Kevin Conner

Built Lemonade, a governed multi-agent platform that functions as an executive job-transition copilot. The system ingests opportunities, normalizes them against a living candidate profile, scores them for fit, applies hard governance filters, and generates tailored recruiter-facing artifacts (cover letters and resume variants) while preserving full human control.

Rather than a single chatbot, Lemonade operates as a coordinated team of specialized agents: search and ingestion agents… Show more

Built Lemonade, a governed multi-agent platform that functions as an executive job-transition copilot. The system ingests opportunities, normalizes them against a living candidate profile, scores them for fit, applies hard governance filters, and generates tailored recruiter-facing artifacts (cover letters and resume variants) while preserving full human control.

Rather than a single chatbot, Lemonade operates as a coordinated team of specialized agents: search and ingestion agents collect roles, scoring agents evaluate fit against skills and constraints, prioritization agents surface the highest-yield opportunities, artifact agents produce cover letters and job packets, and pipeline agents manage status, aging, follow-ups, and referrals. Agents exchange structured outputs, creating a repeatable operating model instead of ad-hoc prompting.

Governance is core to the design. Exclusion filters, explainable scoring, source attribution, and explicit approval gates ensure agents can recommend, draft, and prepare — but never autonomously apply or reach out. A dashboard maintains workflow state, artifacts, and reasoning for traceability.

The result is a practical agentic product: agents divide the labor, strong controls constrain the automation, and the human remains the final decision-maker. I built it both to improve my own search efficiency and as a working demonstration of responsible agentic system design. Show less

I built NPI Builder, an AI-assisted desktop and web tool for creating the core artifacts of New Product Introduction (PCD, PRD, PDD, SDD/HLD/LLD, Risk Register, Launch Checklist, and Post-Launch Review).

The tool combines structured professional templates with a genuinely agentic workflow. A permanent co-pilot called ProdBot handles natural language requests while dedicated actions like “Propagate to next” automatically synthesize rich context from all upstream documents and draft the… Show more

I built NPI Builder, an AI-assisted desktop and web tool for creating the core artifacts of New Product Introduction (PCD, PRD, PDD, SDD/HLD/LLD, Risk Register, Launch Checklist, and Post-Launch Review).

The tool combines structured professional templates with a genuinely agentic workflow. A permanent co-pilot called ProdBot handles natural language requests while dedicated actions like “Propagate to next” automatically synthesize rich context from all upstream documents and draft the downstream artifact with consistency. Bulk operations (fill empties, full-document regeneration, and cross-document consistency reviews) plus versioned checkpoints keep the human firmly in control while removing the mechanical work. The system is LLM-agnostic (works with Grok, OpenAI, or local models via Ollama) and produces clean, stakeholder-ready PDF and DOCX exports.

This project sits alongside a companion Interactive Guide to NPI that maps the full lifecycle, decision gates, RACI ownership, traceability, and common failure modes. Together they reflect how I think about product work: rigorous process married to intelligent automation that respects the craft instead of replacing judgment.

I’m especially interested in Product Management roles and agentic AI product/design or platform work where the goal is to give teams leverage without sacrificing clarity, traceability, or quality.
Show less

Led the evaluation, configuration, implementation, and integration of security tools and procedures to strengthen defensive posture across enterprise environments. Worked hands-on with platforms such as Splunk and Grafana, partnering with cross-functional teams to ensure monitoring, alerting, and security workflows were aligned to best-case operational outcomes.

A key highlight, a project I initiated, was leading the use of Netflix OSS Scumblr to automate security intelligence gathering.… Show more

Led the evaluation, configuration, implementation, and integration of security tools and procedures to strengthen defensive posture across enterprise environments. Worked hands-on with platforms such as Splunk and Grafana, partnering with cross-functional teams to ensure monitoring, alerting, and security workflows were aligned to best-case operational outcomes.

A key highlight, a project I initiated, was leading the use of Netflix OSS Scumblr to automate security intelligence gathering. I architected custom security parameters within the platform to proactively identify potential exposure paths, including sensitive legacy data that remained discoverable across the Internet.

This effort helped surface areas of risk, improve visibility into outdated data-retention practices, and drive updated procedures for identifying, removing, and reducing legacy data exposure.
Show less

I bridge the gap between technical acumen and artistic expression. Inspired by the collaborative culture at Daybreak Game Company (formerly SOE), I dedicated time to refining my graphic design skillset. This project resulted in the creation of a distinct visual style applied to a series of fan art for the title H1Z1.

I worked with and simultaneously increased my knowledge of MVC (PHP) frameworks while coding new functionality and managing graphical and video assets to present to the client. I also performed user acceptance testing to find and report detailed issues with prior development.

Developed a script to detect and remediate the Bash “Shellshock” vulnerability across affected systems, aligning detection logic to the six related CVEs identified at the time.

The script enabled rapid scanning, identification, and patch validation from a single execution path, allowing vulnerable systems to be found and addressed quickly. This reduced manual effort, accelerated remediation, and helped close a critical security exposure immediately.