Who we are:
 
AI Fund is Andrew Ng's venture studio. We identify high-potential applied AI ideas, recruit exceptional builders, and create new companies from scratch. Visiting Engineers join for a 12-week residency in Mountain View to take an idea from concept to working product alongside Andrew and the AI Fund team. If the idea validates, you become a co-founder. We are backed by a $390M fund from top-tier global investors. Our purpose is to build AI companies that move humanity forward.
 
The Idea: AICoS (AI Chief of Staff)
 
AI-powered chief of staff that does the post-meeting work managers hate.
 
Managers spend 15-20% of their week on coordination overhead. After every meeting, someone has to extract action items, sort them to the right project and person, chase follow-ups, prep for 1:1s, and update task boards. The tools exist (Asana, Notion, Slack) but none of them close the loop automatically. Work falls through cracks constantly.
 
LLMs can now reliably extract structured action items from meeting transcripts, understand project context, and generate follow-up communications. The unlock is going beyond capture (which transcription tools already do) to automatic sorting, assignment, and nudging. Think of it as Cursor for task management: it understands your work graph, not just your text.
 
The validation question is whether auto-sorting (not just flat-list capture) is the feature that makes a manager switch tools and use this daily before any team adoption.


What you will do:

Build the meeting-to-action-item extraction pipeline with accurate project and person routing

Develop the auto-sorting engine that maps action items to the correct project context and assignee

Create the nudge and follow-up generation system that drafts communications without human intervention

Ship a working prototype that a single manager can use daily before any team rollout

Validate whether single-player value (one manager, before team adoption) is sufficient to drive retention

Make key decisions around infrastructure and platform choices, balancing prototyping speed with long-term scalability

Apply GenAI building blocks (prompt engineering, agentic frameworks, evals, and guardrails) to real-world development

Collaborate closely with Andrew Ng and the AI Fund team to bring this idea from concept to working product

What you must bring:

Demonstrated experience building applications that incorporate Generative AI

Direct experience with workflow automation systems or productivity tool development

NLP pipeline experience for structured extraction from unstructured text (meeting transcripts, emails, documents)

Integration development across workplace tools (Slack, Asana, Notion, Google Workspace, calendar APIs)

LLM application development with prompt engineering and output quality evaluation

Proficiency with GenAI tools and frameworks: RAG, vector databases, agentic platforms (e.g., LangGraph, CrewAI), LLM evals, and guardrails

Strong back-end skills, particularly with Python, and experience with front-end technologies and modern frameworks

Hands-on experience using AI-assisted coding tools (e.g., Cursor, Claude Code) and understanding of best practices for effective use

Proven track record of architecting, implementing, and deploying scalable AI-powered systems

Ability to write scrappy, disposable code for fast prototyping, and clean code for scalable products

Strong communication skills and ability to work collaboratively across disciplines

Nice to have:

Experience with meeting transcription or audio processing pipelines

Prior work on task management, project management, or calendar intelligence tools

Experience gathering user feedback directly and iterating based on insights

Exposure to rapid prototyping methodologies and shipping MVPs

Contributions to open-source GenAI projects or tools

Prior experience as a founding engineer or technical lead at an early-stage company

Knowledge of advanced architecture patterns (microservices, event-driven architectures, datastore design)

Interest or experience in product design and product thinking