AI and full stack developer

• Natural language processing in Python to curate and clean datasets of human writing
• Fine-tuning large language models
• Experimenting with training parameters
• Developing metrics and eval sets
• Backend development in Python with Django
• Frontend development with React

If you want to work on unsolved AI problems that will revolutionize all digital writing, then you should join our small, top-notch team. We’re looking for early stage, hungry teammates. You should love playing around with AI, maybe from your own projects or past work. But no formal training or number of years of experience is required.

Lots of ownership early. You’ll get to work across our whole stack solving problems in our AI layer, backend, and frontend.

We are a high velocity team and care more about great architecture and software design than leet coding. AI coding (e.g., Cursor, Claude Code) is encouraged for low-level tasks, with detailed human review. We’re looking for teammates who want to work hard and WIN., * Experience fine-tuning large language models (personal projects or past work)

• Experience with Python
• Experience working on a team
• No experience with React is required
• No specific number of years is required, we care more about learning speed and passion for making the best AI writing models

In-Office or Remote Hiring Remotely in United States 80K-120K Annually Entry level In-Office or Remote Hiring Remotely in United States 80K-120K Annually Entry level Work across AI, backend, and frontend: curate and clean writing datasets, fine-tune large language models, run training experiments, develop metrics and eval sets, build backend services in Python/Django, and implement frontend features with React. The summary above was generated by AI

Vaero is a San Francisco-based startup backed by top investors like Y Combinator and Rebel Fund. We’re doing proprietary AI research, NOT an AI-wrapper. We deploy fine-tuned models to write exactly like you.

We’ve invented proprietary techniques (dataset curation, hyper-parameters, custom infrastructure) to train AI models to match any user or brand voice. Our metrics for style matching are miles ahead of even the foundation model companies. And our models pass all AI detectors, such as GPTZero and Turnitin, unlike OpenAI, Claude, and Gemini.

We have a 0.01% team. You’ll work directly with our CEO who was a Computer Science PhD candidate at Carnegie Mellon (his advisor Luis von Ahn founded Duolingo), and went to Harvard Law after getting a perfect score on his LSAT (99.98 percentile). He’s a 2x startup founder, both startups backed by Y Combinator (1.5% acceptance rate).