Staff Software Engineer, Generative AI

Role

Intrinsic Flowstate (www.intrinsic.ai/flowstate) is the primary interface to the Intrinsic platform. External developers use Flowstate to create, deploy and maintain end-to-end robotics solutions. Flowstate is primarily an authoring tool; creating or editing a Flowstate solution includes authoring digital workcells, robot processes, robot skills, and hardware configurations. The Intrinsic R&D team is exploring the use of AI to generate robotics solutions from natural language, multi-modal conversational prompts. These novel workflows can significantly improve the accessibility of Flowstate, which will enable more users to deploy industrial robotics for manufacturing tasks. Before launching these workflows to external users, we need a reliable yet innovative software stack to evaluate, fine-tune, and serve LLMs. 

How your work moves the mission forward

You will lead a greenfield technical area in Intrinsic R&D, and your work will be necessary for deploying generative AI workflows externally. This role comes with a broad scope; your technical leadership will be required to identify the critical projects and to deliver them in phases. If successful, you will make core contributions underlying many workflows of robot software development at Intrinsic, both internal and external.

Your day-to-day work may involve planning and developing workflows to:

• Distill and fine-tune, 
• fine-tune rapidly on user-provided examples of a particular type of data
• distill model down to run on a single machine
• Evaluate 
• develop scalable tools to measure quality of LLMs
• estimate and reduce cost for serving and fine-tuning LLMs
• Serve 
• explore and implement infrastructure to reduce latency and serving costs
  (for example, serve Gemma in a GKE cluster or Gemini Nano in a web browser)

You will develop all the above in the context of authoring robotic solutions, in a user experience built for people with a wide range of robotics knowledge and software abilities. You will work closely with a sub-team within the R&D division, but will also frequently collaborate with other engineering, product, and UX groups within Intrinsic 

Skills you will need to be successful

• Proven success launching and iterating on machine learning or AI workflows and tools in production 
• including large-scale (i.e. distributed/cloud) data processing
• leading large technical projects with multiple stakeholders
• making practical trade-offs and phased deliverables 
• deep knowledge of, and experience with, the state-of-the-art in at least two of the following: 
• fine-tuning (supervised fine-tuning, parameter-efficient fine-tuning, e.g. LoRA),
• serving (self-hosted LLMs for low cost and latency, token optimization for lower costs),
• evaluating (measure quality of fine-tuned LLMs).
• PhD in Computer Science or equivalent work experience

Skills that will differentiate your candidacy

• Strong interest in improving the accessibility of robotics, especially for automated manufacturing
• General knowledge of (and curiosity about) training and deploying foundation models 
• Experience with privacy-preserving data processing
• Experience using LLMs for structured content generation  
• Experience with token optimization for minimizing compute costs of LLMs 
• Startup (or similar) experience --- building the 0 to 1 in a rapidly changing technical landscape 

The US base salary range for this full-time position is between $196,800 - $278,375 + bonus + equity + benefits. Your recruiter will share more about the specific salary range for your targeted location during the hiring process.