Forward Deployed Machine Learning Engineer

The Opportunity

We are looking for a Forward Deployed Machine Learning Engineer with 3+ years of experience to be the first MLE dedicated to the company's Benchmarks and Evaluations vertical. You'll work directly with the GM and researchers to build the technical foundation for AI model evaluation — a critical capability at one of the most exciting data infrastructure companies in AI, backed by a16z, CRV, and others.

What You'll Be Doing

• Partner with the GM and early customers to define, design, and build benchmarks and evals across different domains and modalities
• Build and own the backend infrastructure: data pipelines, execution environments, storage, and orchestration
• Stand up sandboxed environments for agentic evals requiring tools, code execution, or multi-step tasks
• Own the engineering portion of customer engagements end to end
• Identify repeatable eval patterns and infrastructure gaps to drive product opportunities

Key Requirements

• 4+ years of engineering experience with hands-on ML model evaluation work
• Prior ownership of backend and infrastructure (data pipelines, execution environments)
• Experience building benchmarks, evals, or human data pipelines for LLMs (strongly preferred)
• High ambiguity tolerance and a strong bias to action in fast-moving environments
• Strong written communication skills for customer-facing and cross-functional work

About the Company

The company is building the infrastructure layer for AI training data — solving what is arguably the biggest unmet need in AI today: getting access to the right data to train models. The process of acquiring high-quality training data is time-intensive, incredibly expensive, and often ends in failure. Data holders don't know where to start and are rightly concerned about governance, intellectual property, and security implications. AI companies can spend years finding and negotiating access to the data they need.

The company solves these problems by providing an easy-to-use platform that connects data holders with vetted data users — facilitating the secure, efficient, and privacy-centric exchange of AI training data.

What makes the company unique is that they're not just a data broker. The company operates an internal Data Lab that transforms and enriches datasets specifically for AI training, making them significantly more valuable for frontier AI labs and model builders. They partner with organizations across healthcare, media, and other verticals to provide high-quality, real-world training data.

Key highlights: $65M in total funding — $55M Series A (August 2025) backed by Andreessen Horowitz, CRV, Bloomberg Beta, Footwork, Shaper Capital, Liquid 2 Ventures, Flex Capital, and a $10M Seed round with investors including Adam D'Angelo and SV Angel.

Explosive growth — the company hit its entire 2026 goal by June, growing far faster than expected.

~80 employees and scaling rapidly with a lean, high-trust team of builders.

Operating at massive scale — processing up to a trillion images in a single delivery for customers.

Solving AI's data problem is a generational opportunity. The company that succeeds will be one of the largest in AI — and in tech. The company is positioned to be that company.

Founded in 2024, team size around 51 employees.

About the Team

The company's engineering team is a lean, high-trust group of builders who thrive on ambiguity and move with urgency. The culture is best described as obsessed with solving hard problems and delivering great work — not for the sake of grinding hours, but because the pace of growth demands it. As Shayla put it on the intake call: "We've grown way faster than we expected. We hit our entire 2026 goal by June." That kind of rocket-ship momentum shapes everything about how the team operates.

The engineering org is deliberately structured into two types of engineers: those who own customer outcomes end-to-end (Forward Deployed Engineers), and those who build the core product from the learnings those FDEs surface. FDEs in particular are expected to be generalist engineers with high ownership — people who can scope a problem, build a solution, and walk a customer through it, all in the same week. No two days look the same, and that's by design.

Richard Ho, the hiring manager, describes the ideal team member as someone who "can see around corners" — a founding-engineer type who doesn't wait to be told what to build, but instead identifies the right problem, designs around the risks, and ships. The team has a strong preference for people who have operated in fast-moving, high-ownership environments: early-stage startups with traction, FDE-focused companies, or generalist roles at larger orgs where they punched above their weight.

On pace and intensity: nights and weekends are not uncommon, especially when delivery deadlines are tight. That said, the team is explicit that this is not a 996 culture — it's driven by the volume and ambition of the work, not a mandate to be always-on. Several of the strongest engineers on the team have families and balance it all. The expectation is that you're wired to deliver, not that you sacrifice everything to do it.

Cross-functional collaboration is a core part of the role. FDEs work closely with the GM, the Applied Research / DataLab team, the commercial team, and directly with customers and data partners. You'll be expected to hold your own in a room with AI researchers at frontier labs and walk data partners through complex technical decisions — all while keeping the product roadmap honest. Values in practice: Pass the Loved Ones' Test (integrity), Always Find a Way (resourcefulness), Go Fast and Grow Fast (velocity), Practice Kindness and Candor (direct and respectful communication), Deliver Together (shared ownership), and Own the Outcome / Hone the Craft (pride in the work).

Who thrives here: Engineers who are energized by ambiguity, want to own outcomes rather than a narrow slice, and are genuinely excited about building something that hasn't been built before.

Who doesn't: People looking for a 9-to-5, those who want to work on one long-term product problem indefinitely, or engineers who need a fully defined spec before they can move.

The Product

The company is the infrastructure layer for AI training data — a platform that connects data holders with vetted AI companies to facilitate the secure, efficient, and privacy-centric exchange of high-quality training data.

At its core, the company addresses one of the most critical bottlenecks in AI development: access to the right data. Data holders — across healthcare, media, and other verticals — often lack the tools, governance frameworks, and trusted counterparties to safely monetize their data assets. AI companies, meanwhile, can spend years sourcing and negotiating access to the datasets they need to train frontier models. The company solves both sides of this problem through a single, easy-to-use platform.

What sets the company apart is its internal Data Lab — a research and transformation team that doesn't just broker raw datasets, but actively enriches and prepares data specifically for AI training. This means customers receive data that is significantly more valuable out of the box, purpose-built for the demands of modern model builders and frontier AI labs. The company also operates a growing Benchmarks and Evaluations vertical, positioning the company as a neutral arbiter for the AI market. By helping model builders understand current model performance, identify what data will drive improvement, and verify gains over time, the company powers the full AI development lifecycle — from data acquisition through model validation.

The platform is built for scale. The company has processed up to a trillion images in a single delivery, handling massive volumes of structured and unstructured data across modalities including imaging, video, audio, text, and clinical records. Its infrastructure supports the complex, fast-moving requirements of AI researchers who need reliable, high-quality data against tight deadlines.

Tech Stack

Python, LLMs, Data Pipelines, Agentic Systems, Code Execution Sandboxes, RL Environments, TEE/TREs, Backend Infrastructure, Orchestration

---

Application Questions

1. Walk us through a project where you evaluated ML models end-to-end — what was your approach and what did you own?
2. Describe a backend or infrastructure system you built from scratch — what did you build and what were the key technical decisions you made?
3. Tell us about a time you had to move fast with a lot of ambiguity — what was the situation and how did you handle it?
4. What are your main motivators and what are you looking for in your next role?
5. What excites you about Protege? Assess whether they have the soft skills to be client facing or have client facing experience.
6. What is a path forward to benchmark expert level tasks with a high degree of verification, when the models exceed modal human performance?