Software Engineer - Core Platform

Description

The company is an early-stage AI lab building the next generation of frontier multimodal AI models. It was founded by former DeepMind researchers, including Andrew Dai, who previously led work on Gemini. The team has 20 scientists and engineers and recently raised $55M in seed funding from Striker Ventures, Menlo Ventures, Altimeter Capital, and NVIDIA. The company works on hard artificial intelligence problems and is growing quickly.

The role is a Software Engineer position on the core platform. You will design, scale, and harden the distributed systems that run the company's frontier multimodal models. Model progress depends on how well these underlying systems perform.

You will build the infrastructure that links research to production, covering distributed training and evaluation as well as low-latency inference for visual reasoning. You will work closely with researchers so they can focus on model capabilities instead of infrastructure.

Responsibilities

• Build and own high-throughput, low-latency services, asynchronous job execution engines, and streaming pipelines for training, evaluation, and inference on large image and video datasets.
• Improve utilization, concurrency, and communication between services across large GPU clusters.
• Design for fault tolerance, automated node recovery, capacity planning, and auto-scaling across multi-cloud and bare-metal environments.
• Own the full operational lifecycle, including CI/CD with canary releases, SLOs and SLIs, runbooks, and blameless on-call.
• Build benchmarking harnesses that simulate 10-100x load spikes, then fix the concurrency, network, and memory bottlenecks they reveal.
• Work with research scientists to turn new visual reasoning capabilities into stable, ergonomic APIs and internal research platforms.

Requirements

• Have 3+ years of experience designing, shipping, and operating distributed backend systems in production at high data volume and traffic.
• Have strong systems fundamentals in concurrency, networking, memory management, event-driven architecture, and fault-tolerant design.
• Are strongly proficient in C++ and Python.
• Have run production services, including on-call work, incident response, runbooks, and solid telemetry and alerting.
• Have hands-on experience with infrastructure as code (such as Terraform), containers, and orchestrators such as Kubernetes, Slurm, or Ray.
• Have high agency, are comfortable with ambiguity, and can balance speed against long-term stability.
• Have experience at a frontier AI lab, research institute, or a platform team supporting foundation models.
• Have experience building data pipelines or serving systems for multimodal models.
• Are familiar with PyTorch and Rust, and with serving or training runtimes such as vLLM.
• Have exposure to accelerator networking (InfiniBand, RoCE), distributed file systems, multi-tenant isolation, or usage metering.

Conditions

This is a full-time position.

Pay ranges from $300k to $450k, plus a meaningful equity stake and full health and wellness benefits.

Benefits

• A meaningful equity stake in the company.
• Full health and wellness benefits.