Full-Time
Untether AI
51-200 employees
Enhances AI inference with at-memory computing
No salary listed
Senior
Toronto, ON, Canada + 1 more
More locations: Kitchener, ON, Canada
Hybrid position, requiring some in-office presence.
- Computer Science, Engineering or related degree
- 5+ years experience, with a healthy amount of both deep learning and software engineering (2+ years each)
- Experience with one or more deep learning frameworks (preferably PyTorch and/or ONNX)
- Experience working with one or more of the following: CNNs, LSTMs/RNNs, Transformers
- Experience reproducing academic work from the field of AI
- Strong math background
- Strong Python skills
- Experience with C/C++ a plus
- Building tools for our customers to ingest, optimize, and quantize their networks
- Working with deep learning frameworks at a deep level, modifying and extending backend functionality
- Supporting the development of the SDK to produce highly optimized, accurate networks running on our chip
- Masters or PhD in Computer Science, Engineering, or related field
- Embedded or low-level programming experience
- CUDA/OpenGL experience
- Experience deploying neural networks to production
- Experience with model compression techniques such as quantization, quantization-aware training, knowledge distillation, pruning, etc.
Untether AI enhances the speed and efficiency of AI inference workloads using at-memory computing. This method places computing elements next to memory cells, which boosts compute density and accelerates AI inference for various applications like vision, natural language processing, and recommendation systems. The company targets businesses that rely on AI technologies and offers high-performance computing devices, including the runAI200® and tsunAImi® accelerator cards. The tsunAImi® card, for example, provides over 2 PetaOps of performance, making it suitable for demanding AI tasks. Untether AI differentiates itself with its unique at-memory architecture, which allows for more processing power in a compact form factor. The company's goal is to help businesses optimize their AI workloads by providing efficient and cost-effective computing solutions.
Company Size
51-200
Company Stage
Series B
Total Funding
$148.7M
Headquarters
Toronto, Canada
Founded
2018
Simplify's Take
What believers are saying
- Collaboration with Arm targets energy-efficient solutions for automotive applications.
- $20 million funding supports development of machine learning inferencing hardware.
- Partnership with J-Squared opens opportunities in defense and commercial sectors.
What critics are saying
- Emerging competition in energy-efficient AI hardware could threaten market position.
- Rapid AI model evolution may outpace current hardware capabilities.
- Supply chain vulnerabilities in semiconductor components could impact production timelines.
What makes Untether AI unique
- Untether AI specializes in at-memory computing for AI inference workloads.
- Their tsunAImi® accelerator cards deliver over 2 PetaOps per card.
- The imAIgine SDK enhances developer velocity with flexible kernels for neural networks.
Help us improve and share your feedback! Did you find this helpful?
Benefits
Paid Vacation
Health Insurance
Unlimited Paid Time Off
Stock Options
Growth & Insights and Company News
6 month growth
↑ 4%1 year growth
↑ 3%2 year growth
↑ 4%TORONTO--(BUSINESS WIRE)--Untether AI, the leader in energy-centric AI inference acceleration, today announced a collaboration with Arm to develop best-in-class solutions optimized for next-generation automotive applications. As part of this collaboration, Untether AI will enable its inference acceleration technology to be implemented alongside the latest-generation Automotive Enhanced (AE) technology from Arm, to create solutions that address performance, power, safety, and security requirements for advanced Driver Assistance Systems (ADAS) and autonomous vehicle (AV) applications. Compute workloads radically changed in the past several years with the deployment of AI and the increasing complexity of AI and conventional software algorithms has elevated the demand for heterogeneous computing platforms employing domain-specific architectures. This trend, which is especially true in the automotive sector, results in an ideal computing platform that combines a mix of CPUs, AI accelerators, memory, and networking technologies that have been optimized for neural network acceleration and software-defined vehicles (SDV). “The increase in advanced user experiences, ADAS, and AV applications means more heterogeneous compute is needed in the vehicle,” said Suraj Gajendra, vice president of products and solutions, Automotive Line of Business, Arm. “By collaborating with Untether AI to ensure our leading-edge AE technology can work well alongside Untether AI’s acceleration capability, we are enabling our partners to power cutting edge AI workloads across the vehicle.”
Untether says its energy efficient architecture is needed to support next-gen vehicles
"J-Squared Technologies is proud to partner with Untether AI to deliver a solution that meets the rigorous demands of both defense and commercial applications," said Andrew Woollard, Chief Technical Officer at J-Squared Technologies.
Untether AI raises $20 million to develop machine learning inferencing hardware_best bank stocks to buy in india.
To date, Untether AI has raised more than $200 million CAD ($152 million USD) in funding from a group that includes Intel, Tracker Capital, the Canada Pension Plan Investment Board, and Radical Ventures.