Founding Applied Research Engineer

About Rox Rox is building the AI-native revenue operating system. Most enterprise software was built for humans operating dashboards. Rox is built for agents operating systems. Instead of static workflows, Rox runs continuous decision loops powered by real-time context from across the enterprise. Agents analyze signals, reason about them, and take action — automatically. To make that possible, we are building infrastructure that combines: Distributed systems Real-time decisioning Agent execution frameworks Semantic retrieval systems Low-latency streaming infrastructure Knowledge graph architectures We're backed by Sequoia, GV, and General Catalyst and have grown rapidly from 15 to 75 people while shipping our platform years faster than the market expected. We already support Global 2000 customers across banking, hardware, construction, and sovereign AI. We are building for what enterprise AI systems will require two years from now — not just what exists today. Why This Role Exists There is a gap in the world right now between what research produces and what production demands. Benchmarks get gamed. Evaluations miss what actually matters. Models pass tests in sandboxes and fail in live environments where the data is dirty, the state is changing, and the cost of being wrong is real. We are in market. We are running agents against enterprise data at scale, every day. We see exactly where the frontier of research meets the floor of production — and we are building a team to close that distance permanently. The Applied Research team at Rox does not publish for the sake of publishing. It researches because the problems we are facing in production require it. What This Team Works On The problems we care most about right now: Agent reliability at scale. How do you know an autonomous agent is doing the right thing when it's executing thousands of actions per day across real enterprise data? Evaluation infrastructure, behavioral benchmarking, and reliability frameworks for long-horizon agent workflows. Self-healing agents. Agents that detect their own failure modes, adapt their behavior, and recover without human intervention. Not in a demo — in production, under SLA, against messy real-world data. Knowledge graph and context engineering. We made a bet three years ago that the enterprise would push all of its data into the warehouse. We were right. Now we are building the infrastructure that turns static warehouses into semantic graphs that agents can reason over in real time. Memory, retrieval, state — these are unsolved problems at our scale. Evaluation for full trajectories, not just outputs. Most eval infrastructure measures end results. We care about the path. How do you design evaluation systems that can reason about the quality of a decision chain, not just whether the final answer was right? Latency and reliability at the inference layer. When an agent is operating in a live revenue environment, it cannot wait. We are building retrieval and inference systems that operate at latencies most people treat as aspirational. What You'll Do Design and run research programs directly tied to production agent problems Build evaluation frameworks that measure what actually matters in live systems Work on agent memory, retrieval, and context systems alongside the Core and Platform Eng teams Translate research findings into infrastructure improvements with measurable production impact Help define Rox's long-term research agenda as we scale Collaborate closely with engineering leads who have built at the frontier — our team includes IMO, IOI, and ICPC medalists, and researchers from DeepMind, OpenAI, and top academic labs What We're Looking For You might be a fit if: You have spent real time thinking about how agents fail in practice, not just how they perform on benchmarks You have built evaluation systems and know exactly where the standard approaches break down You can write code well enough to implement your own ideas and run your own experiments — and ship things that make it into production You move fast. Research at Rox is not slow. The environment changes monthly and the team ships continuously. You are more interested in problems that matter than problems that are easy to publish Particularly relevant backgrounds: Agent evaluation and behavioral benchmarking Retrieval-augmented generation, semantic search, and knowledge graph systems RL applied to real-world agent behavior Production ML systems — latency, reliability, observability Data pipeline and agent orchestration infrastructure Post-training, fine-tuning, and model adaptation for production use cases A PhD is not required. Strong research instincts and the ability to ship are. What Success Looks Like In your first few weeks: You understand Rox's architecture, the problems we're actively fighting in production, and where the research gaps are. You have opinions and you share them. In your first few months: You are running experiments that directly inform how we build. Something you worked on is in production. Over time: You are defining the research agenda for one of the most interesting applied AI problems in the enterprise — and building systems that no one else has built before. Why Join Now We are at an unusual moment. We are large enough to have real scale, real customers, and real production problems that create genuinely interesting research questions. We are small enough that you will be one of a handful of people shaping what the Applied Research function looks like and what it prioritizes. The team is exceptional. The problems are hard. The impact is real and measurable. If you want to work on applied research where the feedback loop is a live enterprise system and not a leaderboard — we'd love to talk. Location This role is based onsite in San Francisco. We strongly believe that the speed and complexity of the problems we're solving benefit from in-person collaboration. We actively relocate exceptional engineers and researchers from around the world and currently have team members who have moved from Europe, Canada, Asia, and Australia to join Rox.