AI Product Engineer

Fireworks is the platform for specialized intelligence, enabling companies to build, train, and serve AI models tailored to their own data, workflows, and products. Founded by the team behind PyTorch and backed by AMD, Atreides, Benchmark Capital, Index Ventures, Lightspeed, NVIDIA, Sequoia Capital, and TCV, Fireworks powers production AI with hundreds of state-of-the-art open models across text, image, embedding, audio, and multimodal workloads. Today, Fireworks is a Series D company valued at $17.5 billion, bringing together an ambitious, collaborative team that's building the future of enterprise AI.

We just shipped Fireworks Nexus : a drop-in platform that lets engineering organizations route their AI workloads off expensive proprietary models and onto high-performing open models like Kimi-K3 and GLM-5.2, without changing how a single developer works. The early numbers are the kind that change how an industry buys intelligence: 3–5x lower AI spend, roughly a third off the cost of every merged PR, and a blended token rate about a quarter of the closed labs , all while matching or beating frontier models on real engineering work from real customer repositories. The context is simple. AI spend at serious engineering orgs is exploding, one company burned its entire annual AI budget in four months, and most of that money goes to running routine work at frontier prices. Open models just crossed the intelligence-cost curve. Nexus is how teams capture that, and we're building it in the open: FireConnect ships under Apache 2.0, installs in one line, and plugs straight into Claude Code, Codex, and OpenCode. We're looking for a product-minded engineer to join as a Member of Technical Staff and own the surfaces that make this real: the intelligent router that scores and directs every request, the cost-observability and enterprise-control layers that give an org visibility over its entire AI footprint, the FireConnect integrations that make adoption a single command, and the core platform underneath; inference APIs, the developer console, and fine-tuning workflows. This is a high-autonomy IC role on a small team building something that ships to production the day you write it, at a scale almost nowhere else can offer. You'll own features end to end — from understanding the problem, to designing the solution, to shipping it, to watching how developers actually use it. If you're the kind of engineer who ships things other people think are impossible, on timelines other people think are impossible, this is a place where that has enormous leverage and nothing standing between you and impact.

Own product features end-to-end across Nexus and the broader Fireworks platform, the intelligent router, FireConnect harness integrations, cost observability, enterprise controls and policy, the developer console, API surfaces, the model playground, fine-tuning workflows, and billing, scoping the problem, building it, shipping it, instrumenting it, and iterating on real usage. Build the routing and migration layer that scores request difficulty in real time and sends routine work to open models while passing hard tasks through to a customer's existing provider, the system that delivers 3–5x cost reductions without sacrificing quality. Make adoption frictionless: extend FireConnect and our Anthropic- and OpenAI-compatible APIs so a team can drop Nexus into their existing tools with just a base URL and a model ID, and improve the caching and infra paths that turn that into material cost savings. Work across the full stack, frontend (React/TypeScript), backend services (Python, Go), APIs, and data layers, to ship polished, production-quality product surfaces for deeply technical users. Build for products evolving week to week: new models, new serving paths, new consumption models, new enterprise controls, and new multi-agent harness tooling. Partner closely with Infrastructure, Applied Research, Growth, Data,