Level: Mid-Senior
Salary: $180k-280k
Stock Options

Are you passionate about building cutting-edge machine learning models and applications that can generate, enhance, and design data? Do you want to join a dynamic and growing team of ML experts, engineers, and designers who are pushing the boundaries of creativity? If so, we wrote this paragraph with GPT so don’t worry we’re not that corporate!

We are looking for a machine learning generalist for our UI-AI team. We are building generative ML models that help designers create their best work by combining amazing design context, powerful models, and curated datasets. Your expertise will help us take the design data and ideas we have and turn them into shipped models deployed to thousands then millions of designers worldwide.

We want to create an environment with a really tight idea → shipping pipeline. We don’t think we can out-research big companies but we can put designers, ML research, and engineering in the same room and create something really useful today.

Imagine running multiple training runs on a diffusion model in a week and picking one to ship to our beta testers, with a monthly cadence of larger improvements. Help us decide which model is working better in production, work to combine user signals with generated data to iterate on our text→image, layer→layer &c models, and help us define the future of design AI!

Your day to day will be debugging issues with training and eval code, running experiments, and working with product teams to get more out of our ML models (UI-AI) in production.

Our Data Stack

• python

• pytorch

  • diffusers, &c

• weights & biases

• 🤗 hub

• data processing with modal

• postgresql, mysql / planetscale

• S3

• Figma Plugins, APIs, etc 🤷

• Your favorite tool here

Models

You might be working with these this month, maybe something completely different in the future

• diffusion models

• transformer language model fine-tuning

• multi-modal generation

Technical

• 3+ years of python, julia, or lua

• 3-5+ years of data science or ML eng, industry experience

• familiar with training, inference, and evaluation

• experience with current ML workflows in pytorch, jax, or tensorflow

• experience deploying ML models in production

• experience collaborating with others on ML (either open source or internal)

• experience tracking and evaluating experiments

• strong SQL and practical unix skills

Nice to have

• incorporates AI tools as much as possible into workflow, eg copilot, ChatGPT, &c

• experience managing datasets > 100GB

• familiar with Apache Arrow or similar tools

• experience doing web scraping at scale

• experience optimizing model speed with XLA / Triton / Flax / DeepSpeed or other accelerator compilers

• experience using kubernetes or cluster management tools

• basic web dev skills, eg react typescript

Theoretical ML Skills

• You've dealt with the whole stack, talking to customers, iterating on experiment design, models and hypotheses and testing them

• Interested in making a contribution to the broader design ML community

  • (optional) author on research papers

  • (optional) participated in open collaborations in industry or academia

Responsibilities

• work with rest of UI-AI team and product teams to define custom ML models

  • good technical communication skills (external & internal, technical & non-technical). explain what you’re doing and why.

• calculating and tracking existing metrics like CLIP-distance and Inception score

• creating new metrics relevant to our tasks

• collab w/ others in ML community

Creative

• Basic understanding of design methods & tools

  • Basic understanding of typography / color / icons

  • Some idea about how frontend HTML & CSS works

• or, a high curiosity and willingness to learn

• Desire to dig in and learn about design as practiced today and help define its future

Opportunities for growth

• Grow into a team / tech lead at a growing startup

• Participate in the nascent community of generative AI for design

• Published or publicly shipped results multiple times a year

• Build SOTA models for design-specific tasks

Application Process

• click “apply” on our read.cv page

• quick chat with someone on the team

• skills assessment eg

  • quick questions related to training ML models

  • explain how you solved X in the past)

• paid take-home project