2
Founding AI Engineer at Harper | Y Combinator
www.ycombinator.comThe Mission
We're building the first truly intelligent insurance brokerage, transforming a $100B+ Excess & Surplus(E&S) insurance market that runs on pre-internet systems. We are looking for exceptional engineers who can architect and deploy AI systems that make complex decisions at unprecedented speed. This isn't about adding chatbots to insurance - it's about building AI systems that can handle intricate, multi-step workflows autonomously, turning weeks of manual work into automated, instant processes.
You're Our Person If
You ship features daily and think in weekly roadmaps, not months
You're obsessed with velocity - if something can be built in a week, you'll do it in days
You think in first principles and build complex systems in one-third the normal time
You're comfortable throwing everything away and starting from scratch when needed
You can look at complex workflows that "experts" say can't be improved, see through to the core problems, and build solutions at a speed that makes others uncomfortable
You've architected complex AI applications from the ground up
You have strong opinions about AI infrastructure and aren't afraid to defend them
You judge systems by their real-world impact, not their theoretical elegance
You think deeply about AI safety and reliability in production systems
Hard Requirements
Must be in San Francisco during YC W25 batch (Jan-March 2025)
Long term: Based in San Francisco or New York
Deep experience with modern AI/ML stack
Strong track record of shipping production AI systems
Comfort with rapid iteration and constant change
Indicative Timeline Targets
Our pace is ambitious because we know what's possible. Here's what your impact could look like:
First Few Days: Master our data flows while simultaneously shipping your first feature - an AI form generation system with RAG capabilities. You'll be integrating Elasticsearch for intelligent document retrieval, proving we can move at unprecedented speed.
First Week: Transform how we handle customer interactions by deploying real-time voice AI that works alongside our human operators. This isn't just about transcription - it's about building AI that can understand complex insurance conversations and make intelligent decisions in real-time.
Second Week: Build intelligence into our market communications. Create systems that prioritize and route submissions to carriers, monitor for objections, and manage complex data flows - turning what typically takes weeks into an automated, instant process.
Third & Fourth Weeks: Take our AI systems to the next level - from handling complex voice conversations to deploying autonomous agents for payment processing and portal interactions. You'll build systems that can navigate the maze of insurance websites, extract critical information, and feed it back into our workflows.
Your Domain & Impact
You'll be building our core AI infrastructure:
Decision Systems: Design AI that can handle complex insurance decisions
Workflow Automation: Build systems that orchestrate human-AI collaboration
Infrastructure: Create robust, scalable AI platforms that can evolve rapidly
Tooling: Develop proprietary tools for insurance-specific AI challenges
Evolving Tech Stack
Building for velocity today, engineering for scale tomorrow:
Data & Pipelines:
Unstructured, Airbyte
Retool workflows, make.com (http://make.com), Zapier
Proprietary fine-tuning data compilation workflows
AI/ML Infrastructure:
Embedding Models: OpenAI, HuggingFace
Vector Search: ElasticSearch, Exa.ai (http://Exa.ai)
Development: Vercel AI Playground, abacus.ai (http://abacus.ai)
Orchestration & Operations:
Retool workflows, Temporal workflows
Custom wrapped LangChain and LlamaIndex
Bland.ai, Braintrust.dev (http://Braintrust.dev)
Weights & Biases, Langsmith
LLM Stack:
APIs: Claude, OpenAI, Groq
Hosting: Hugging Face, Together.ai (http://Together.ai), modal.com (http://modal.com)
Caching: Vercel KV Logging, Claude caching
Platform & Frontend:
Cloud: GCP
Deployment: Vercel
Interface: Retool
CRM: HubSpot
Communications: OpenPhone
How We Execute
Make Requirements Less Dumb: Question everything. If someone says "that's how insurance works," that's exactly where we start
Delete the Process Step: The fastest process is no process. Every step must justify its existence
Optimize: Make it smarter. If AI can do it better, it should
Accelerate: Speed compounds. What can be done in a day should be done in an hour
Automate: Transform human expertise into AI systems that scale
Our SDLC
Boundaries of Separation: Build core IP, leverage existing tools for speed
Micro Applications: Rapid prototyping with bolt.new (http://bolt.new), Replit agents
AI-Assisted Development: Cursor for complex builds
Automated Maintenance: Considering Devin for enhancements and maintenance
Weekly Releases: Major features ship weekly, minor features ship daily
Our Philosophy
"Move fast. Break things. Fix them faster."
"Strong opinions, loosely held"
Join as an Owner, Not an Employee
Base compensation for this role ranges from $100-180K depending on your experience:
Early Career (1-2 years): $100-125K
Mid-Level (3-5 years): $125-150K
Senior (6+ years): $150-180K
But let's be clear - if you're just comparing base salaries, you're missing the point.
We're building something that could transform a $100B+ industry, and we want our early team members to have meaningful ownership in that journey. That's why we offer significant equity packages that make you a true owner, not just an employee. We're looking for people who understand that the real upside comes from building and owning something revolutionary, not from maximizing their base salary.
Think like an owner because you will be one. If you're primarily optimizing for cash compensation, we're probably not the right fit. If you're excited about building something massive and owning a meaningful piece of it, let's talk.
You must log in or register to comment.
