Perforated is a public participant in Nebius’s inaugural AI Lift research cohort. The eight-week pilot provides each team with Nebius compute, hands-on support from solutions architects, go-to-market workshops and introductions to venture investors.
Perforated develops learning methods inspired by biological dendrites to build smaller, more efficient neural networks. Its software integrates into existing PyTorch training workflows and adds neuron-specific learning signals. In the cohort announcement, founder Rorry Brenner describes the project’s goal of developing a small, high-performance language model that could enable local applications on resource-constrained devices such as phones and laptops.
Nebius names Perforated AI as a public AI Lift research-cohort participant. Each team receives Nebius compute and engineering support during the eight-week pilot. Perforated’s founder describes a small language model for resource-constrained devices as the project’s intended outcome.
The Lab Gets a Lift12 Oct 2026 · Nebius Science · Web pageNebius names Perforated AI as a public AI Lift research-cohort participant. Each team receives Nebius compute and engineering support during the eight-week pilot. Perforated’s founder describes a small language model for resource-constrained devices as the project’s intended outcome.
Perforated, operated by Perforated AI, Inc. in Pittsburgh, develops software that adds artificial dendrites and neuron-specific learning signals to neural networks during training. It serves machine-learning engineers seeking better accuracy with fewer labeled examples and smaller models for cloud and edge deployment. Its offerings include an open-source PyTorch library and the commercial Perforated Suite, which includes its Perforated Backpropagation training algorithm and Perforated Studio interface for configuring runs and inspecting results. Optimized models export into existing deployment workflows without requiring a Perforated runtime.
Perforated discloses $1.11 million raised in ongoing offering
Perforated disclosed $1,114,500 sold to nine investors in a $4 million offering involving SAFEs. The September filing lists the first sale as May 11, 2026 and $2,885,500 remaining to be sold.
Perforated appointed Dean Alderucci as chief technology officer, effective immediately. He will oversee technology strategy, infrastructure, engineering and technical implementation.
Perforated develops dendritic learning methods and software to improve neural-network accuracy and parameter efficiency. · Model developers
How they use Nebius AI Cloud
Perforated is a public participant in Nebius’s inaugural AI Lift research cohort. The eight-week pilot provides each team with Nebius compute, hands-on support from solutions architects, go-to-market workshops and introductions to venture investors.
Perforated develops learning methods inspired by biological dendrites to build smaller, more efficient neural networks. Its software integrates into existing PyTorch training workflows and adds neuron-specific learning signals. In the cohort announcement, founder Rorry Brenner describes the project’s goal of developing a small, high-performance language model that could enable local applications on resource-constrained devices such as phones and laptops.
Nebius names Perforated AI as a public AI Lift research-cohort participant. Each team receives Nebius compute and engineering support during the eight-week pilot. Perforated’s founder describes a small language model for resource-constrained devices as the project’s intended outcome.
The Lab Gets a Lift12 Oct 2026 · Nebius Science · Web pageNebius names Perforated AI as a public AI Lift research-cohort participant. Each team receives Nebius compute and engineering support during the eight-week pilot. Perforated’s founder describes a small language model for resource-constrained devices as the project’s intended outcome.
Perforated, operated by Perforated AI, Inc. in Pittsburgh, develops software that adds artificial dendrites and neuron-specific learning signals to neural networks during training. It serves machine-learning engineers seeking better accuracy with fewer labeled examples and smaller models for cloud and edge deployment. Its offerings include an open-source PyTorch library and the commercial Perforated Suite, which includes its Perforated Backpropagation training algorithm and Perforated Studio interface for configuring runs and inspecting results. Optimized models export into existing deployment workflows without requiring a Perforated runtime.
Perforated discloses $1.11 million raised in ongoing offering
Perforated disclosed $1,114,500 sold to nine investors in a $4 million offering involving SAFEs. The September filing lists the first sale as May 11, 2026 and $2,885,500 remaining to be sold.
Perforated appointed Dean Alderucci as chief technology officer, effective immediately. He will oversee technology strategy, infrastructure, engineering and technical implementation.
