Mark Lape
Applied AI portfolio
I build practical AI applications around real business workflows. My work turns notes, emails, documents, and project information into structured, reviewable outputs that help teams make decisions and complete work.
The pattern behind the projects
I focus on the part between a business problem and a usable AI workflow: understanding the process, structuring the information, and keeping people in control of the result.
Business input
AI interpretation
Structured output
Human review
Action and history
Selected evidence
A focused set of internal and product work showing how I approach inputs, model-assisted decisions, structured outputs, and human review.
Internal AI copilot
Captures decisions, blockers, and action items through chat, file uploads, and structured inputs, then generates updates for internal teams, clients, and vendors. The source input, generated output, and project history stay available for review.
Relevant stack: React · TypeScript · Supabase · OpenAI API · PDF, DOCX, and spreadsheet parsing
Multimodal AI assistant
Turns free-form notes, voice input, and attachments into Decisions, Meetings, Commitments, and Artifacts. When intent is unclear, the assistant asks a clarifying question instead of guessing.
Relevant stack: React · TypeScript · Firebase · Firestore · OpenAI GPT-4.1
AI-assisted business workflow
Connects project search, AI-assisted drafting, user editing, image management, and formatted PowerPoint export. It helps proposal teams move from scattered project information to a usable deliverable.
Relevant stack: MySQL · SharePoint · OpenAI API · PptxGenJS
Grounded knowledge assistant
Makes complex regulatory information easier to query. Users receive focused answers with source references and calculations, while administrators can maintain the underlying knowledge base.
Relevant stack: Next.js · React · OpenAI GPT-4.1 · Supabase · Zod
Microsoft Copilot experience
Working range
My strength is shaping the workflow, refining the experience, connecting the services, and testing whether the result is useful to the people who need it.
Strong working tools
OpenAI API · Claude · Codex · Next.js · Supabase · GitHub · Vercel
Working knowledge
Microsoft 365 OAuth · Google APIs · DeepSeek API · Firebase · n8n
Familiar with
Microsoft Graph API · Python · Flask · Docker
Let's build something useful
I'm interested in AI Application Engineer, AI Agent Developer, Generative AI Developer, Copilot, and AI Automation opportunities.