Sophie Zielinskie

LLM-powered manager development simulation running natively in Slack. Managers receive realistic employee scenarios (e.g., burnout, harassment, performance, leave) from an AI-generated persona and respond in their natural workflow. The LLM evaluates responses using full-context behavioral inference — not keyword heuristics. Flags compliance risk and AI-assisted responses automatically. Built solo in one week.

Engineered using Goose and Claude to generate fully branded SCORM 1.2 HTML courses end-to-end — bypassing the manual GUI bottleneck of tools like Articulate Rise. Includes stakeholder review overlay and automated revision workflow. Exploring xAPI/Snowflake integration to unlock behavioral learning signals at scale.

Operationalized org-level OKR into a sustainable program serving Customer Advocates across CustOps.

Composed of formal training, applied experiments, and communities of practice to address an identified skill gap in AI-fluency and human-in-the-loop capabilities. Originally designed for leadership — in progress of expanding enterprise-wide due to relevancy and content quality.

Designed the annual global enterprise sales kickoff. Full program design, content, and delivery coordination.

Launched the Enterprise Pulse survey via CultureAmp across the global Revenue organization.

Engineered a custom JIRA-based training request and ticketing system with automated API integration — bringing structured intake to an org that had none.

One month to launch. Cancelled vendor contract. Built from scratch: full curriculum, 30+ facilitator onboarding, real-time feedback ecosystem with Slack community and custom HTML emails delivering personalized learner quotes to each facilitator by name after every session. Shipped on time.

Designed COVID Return-to-Office training with no existing playbook, 3 months into her Verizon tenure. Applied behavioral design throughout: color psychology (blue over brand red — calm over coercion), empathy mapping, and image transitions from black and white to color to signal normalcy returning. Rather than multiple choice, she built commitment devices — prompting learners to declare specific actions rather than select abstract answers — because behavior change requires agency, not recall.

Designed to reduce preventable catastrophic outages (CATAs) among unionized field technicians and tradespeople. Built the framework Prepare, Pause, Confirm — adopted by executive leadership into company-wide messaging. Developed extensive APIs that diagnosed the underlying human state driving responses: flagging overconfidence, complacency, being unskilled, or miseducated — and separately recognizing expertise-driven confidence, recognition-motivated high performance, and mastery. The system was built to surface that distinction accurately, enabling tailored remediation strategies for each state.

Contributing role within a much larger team effort — not sole ownership. Led ideation and content writing for complex branched scenarios built around learner role, knowledge level, prior experience, and risk profile. Partnered with the xAPI developer to ensure assessment questions were properly aligned to behavioral signals and compliance outcomes. The program incorporated learner-controlled audio — an accessibility and engagement innovation. Extensive xAPI data collection enabled targeted reinforcement for high-risk topics. The program earned multiple awards and became a flagship example of innovative L&D development at Verizon.

Facilitated exec-only sessions across 10+ functions: HR, Legal, Finance, Technology, Network, Retail. Captured executive interviews for social proof in downstream training. Turned live skepticism from senior leaders into in-room proof of the framework's core principle.

Converted global leadership program to virtual without sacrificing quality or outcomes.