Gen AI Framework Engineer

About Us
At traceVision, we turn video (pixel data) into data computers can use. We aim to teach computers to see the way humans do—with persistent memory, tracking, and emotional responses to scenes or actions. We've been pioneering real-world AI for over a decade, and we're cash flow positive—which means we're in control of our own destiny and growing!

Our flagship product, traceup.com, delivers video highlights for parents of youth sports athletes. We're processing 8 billion frames of video daily and creating highlights for nearly 200,000 athletes each week. We've also expanded into security and retail sectors, with public companies building on Core Data across operations, merchandising, asset protection, and internal engineering teams.

We've been in love with AI before it became a buzzword and deployed our first models in production in 2015. AI writes a lot of our code, and we've developed internal frameworks that support our GTM teams (Sales, Customer Success, Support, Operations, Shipping) and engineering teams (code writing, processing supervision, annotations for model training). If you don't believe that AI can be an effective salesperson or can write production code, you won't fit in!

We're a small team that works hard and takes ownership. Every employee here is a single point of failure. This is a high-stress environment, and we often work on weekends. We keep the main thing, the main thing. We're fully remote but meet once a quarter somewhere in the U.S. for onsite collaboration and intense brainstorming.

About the Role
We use a lot of Generative AI inside traceVision. It runs our entire GTM team (Sales, Customer Success, Support, Operations) using an internal framework named TRINA (Trace Realtime Intelligent Network of Agents). It features function calling, adversarial agent structures, Chain of Thought reasoning, and more.

On the coding side, we have an in-house framework named BRAIN that writes 20% of our code. The idea is that an engineer should never start off with a blank page. BRAIN knows about your coding style, the company's coding style, other repos, etc.
Until recently, this has been someone's side project. We need an engineer to take over this project and build on top of it. We're thrilled with what it's done for the business (removed $14M of OpEx) and believe there's much more to do to make our teams superhuman.

A good measure of success for this project is how efficient the company can grow revenue. We measure that by looking at revenue / employee. We're #1 across all our VC's investments, and we think it can climb much higher.

Technologies We Use
Languages: Python, some Node.js
Frameworks and Tools: AWS (Beanstalk, Pinecone, LangChain), OpenAI, Anthropic, Groq, Gemini
Infrastructure: AWS, GCP
Concepts: Function calling, adversarial agent structures, Chain of Thought prompting

What You'll Be Doing
Own and enhance our internal Generative AI frameworks (TRINA and BRAIN).
Refactor and optimize existing codebases to be faster and more efficient.
Implement new features to expand the capabilities of our AI agents.
Collaborate with GTM teams to understand their needs and improve AI integrations.
Integrate BRAIN more deeply into our development workflows.
Convert Python code into compiled languages (Rust, Go, C++) for performance gains.
Stay current with the latest advancements in Generative AI and propose innovative solutions.
Measure and improve the impact of AI on operational efficiency and revenue per employee.

What We're Looking For
Strong proficiency in Python and experience with compiled languages like Rust, Go, or C++.
Hands-on experience with Generative AI models and frameworks (OpenAI, Anthropic, etc.).
Familiarity with AWS services (Beanstalk, lambda, batch), tools like LangChain, and vector databases like Pinecone.
Experience with function calling, adversarial agent structures, and Chain of Thought prompting.
Proven ability to optimize and refactor code for performance.
Self-motivated and comfortable in a high-stress, fast-paced environment.
Willingness to take ownership and work weekends when necessary.
Code/project you can share live during your interview.
Favorite paper on getting the most out of LLM’s
Interesting side projects that showcase your skills and passions—we're going to ask you about them!

Things We Don't Care About
Where you went to school.
What your LeetCode score is.**