Build the future
of visual memory

We're hiring a Senior Machine Learning Engineer to work alongside CTO on our core modeling team. The mandate is threefold: ship models faster, strengthen our system design, and harden the architecture our models run on. You'll own the path from a promising experiment to a deployed, monitored model — and make that path shorter and more repeatable for everyone behind you.

This is a high-ownership role for someone equally comfortable reasoning about model architecture and the infrastructure that trains, serves, and evaluates it.

Partner closely with CTO to take models from research prototype to production, cutting iteration time at every stage of the loop.

Design and improve training, evaluation, and serving infrastructure so experiments run faster and results are reproducible.

Make architecture decisions for our ML systems — data pipelines, model serving, inference optimization — and document the trade-offs.

Build tooling and abstractions that let the team launch, compare, and roll back models with confidence.

Optimize model performance and cost: latency, throughput, memory, and GPU utilization across training and inference.

Establish strong engineering practices — testing, monitoring, CI/CD for models — so "faster" never means "more fragile."

Mentor engineers and raise the bar on technical design across the team.

5+ years of experience building and shipping machine learning systems in production, with ownership across the full ML lifecycle — from data pipelines and model development to deployment, monitoring, and continuous improvement.

Strong software engineering fundamentals — you write clean, tested, maintainable production code and can design reliable systems beyond experimental notebooks.

Deep hands-on experience with modern ML frameworks and infrastructure, including PyTorch (or equivalent), distributed training, model fine-tuning, model serving, and GPU-aware optimization.

Experience improving ML system architecture and engineering efficiency — you've designed or reworked systems that made teams faster, models more reliable, or infrastructure more scalable.

Strong understanding of production ML infrastructure, including containers, orchestration, cloud platforms, observability, and scalable deployment practices.

Experience with transformer-based architectures and modern ML workflows, including working with embedding models, retrieval systems, or multimodal pipelines.

Pragmatic engineering judgment — you know when to optimize, when to ship, and when a simpler solution is the right one.

Ability to thrive in a fast-moving startup environment, taking ownership of ambiguous problems and driving solutions from idea to production.

‍

Nice to Have

Experience building or optimizing multimodal AI systems, especially image/video understanding models or large-scale visual AI applications.

Experience with large-scale retrieval systems, vector databases, embedding pipelines, or memory architectures.

Hands-on experience with inference optimization techniques, including quantization, batching, distillation, memory optimization, and acceleration frameworks such as vLLM, Triton, TensorRT, or similar tools.

Experience optimizing GPU utilization, model latency, and serving efficiency in production environments.

Background scaling ML infrastructure at an early-stage startup or high-growth engineering team.

Contributions to open-source ML infrastructure, tooling, or research projects.

‍

Real ownership over how we build and ship models, working directly with strong engineers like our CTO on problems that are both research-hard and systems-hard. The work you do to make delivery faster and architecture cleaner compounds across the whole team.

We are building the memory layer for visual AI. Our infrastructure runs inside products from NVIDIA, Vivo, Comcast, and Ring, and our research is shaping how the next generation of AI systems remember the real world. Founded by a Cambridge PhD with twenty-eight publications, backed by world-class investors, and partnered with the companies defining frontier AI.

We are building the memory layer for visual AI. Our infrastructure runs inside products from NVIDIA, Vivo, Comcast, and Ring, and our research is shaping how the next generation of AI systems remember the real world. Founded by a Cambridge PhD with twenty-eight publications, backed by world-class investors, and partnered with the companies defining frontier AI.

Visual memory is a research field, not a product feature. We are hiring a research scientist to push the boundaries of how machines see, remember, and reason across time. You will work on the core models that power our platform, from multimodal representation learning to long-horizon temporal reasoning, and publish openly with the rest of the team.

.Long-context video understanding, beyond the minute-scale limits of current foundation models. New methods for multimodal embedding across video, audio, and language. Temporal reasoning systems that connect events, objects, and people across hours or years of footage. Open research that ships into production, used by enterprises operating at scale.

You will work closely with our engineering teams to take research from paper to deployment, and with our partners at NVIDIA, Samsung, Lenovo, and others on real-world problems no one has solved yet.

PhD in computer vision, machine learning, or a related field, or equivalent industry research experience. A track record of publications at top-tier venues (CVPR, ICCV, ECCV, NeurIPS, ICLR). Strong fundamentals in multimodal learning, video representation, or temporal modeling. Comfort with large-scale training infrastructure. A taste for problems that matter outside the lab.

Compensation calibrated to top-tier research labs, including meaningful equity. Full benefits, health, dental, vision, mental health support. Conference and publication budget, no caps on conferences attended. Remote-first with quarterly team gatherings in San Francisco. The chance to build a research field, not just contribute to one.

We are building the memory layer for visual AI. Our infrastructure runs inside products from NVIDIA, Vivo, Comcast, and Ring, and our research is shaping how the next generation of AI systems remember the real world. Founded by a Cambridge PhD with twenty-eight publications, backed by world-class investors, and partnered with the companies defining frontier AI.

We are building the memory layer for visual AI. Our infrastructure runs inside products from NVIDIA, Vivo, Comcast, and Ring, and our research is shaping how the next generation of AI systems remember the real world. Founded by a Cambridge PhD with twenty-eight publications, backed by world-class investors, and partnered with the companies defining frontier AI.