Our verdict
OpenTeams is a strong technical alternative when the assignment involves PyTorch, inference performance, or ownership of an open-source AI environment. Its engineering record is more relevant to that job than a generic workflow-automation offer. The Nebari platform and a changing product-led business model require a clear distinction between custom engineering and a continuing platform relationship.
The right buyer
- Engineering teams optimizing model execution, compilation, or open-source ML infrastructure.
- Enterprises that need an AI environment they can control and operate.
Reasons to look elsewhere
- A simple business chatbot may not require its lower-level engineering expertise.
- Nebari delivery and custom performance work are different purchases with different support obligations.
OpenTeams, At A High Level
OpenTeams brings together open-source engineering and enterprise AI deployment. Its founder and CEO, Travis Oliphant, helped create NumPy and SciPy and previously built Anaconda and Quansight. In May 2025, OpenTeams acquired Quansight’s consulting division, including engineers contributing to PyTorch. The current engineering account describes work on torch.compile, Dynamo, Autograd, and optimized kernels. These are components that affect how models execute, giving the performance-engineering offer a concrete foundation. Nebari supplies an enterprise AI environment alongside custom services. Oliphant resumed the CEO role in March 2026 as the company moved toward a more product-led model. A buyer should establish whether the proposed team is delivering bespoke optimization, a managed environment, or both.
Scores by criterion
| Criterion | Weight | Score | Basis |
|---|---|---|---|
| Production track record | 20% | 8.0 | The engineering team documents continuing contributions to production ML infrastructure. |
| Technical depth | 15% | 9.0 | PyTorch compilation, optimized kernels, Python, and enterprise AI environments. |
| Industry depth | 10% | 7.5 | Technical computing and enterprise AI infrastructure are clear specialties. |
| Delivery model and handoff | 15% | 8.0 | Custom engineering, environment deployment, and support are available, with scope to distinguish. |
| Commercial clarity | 10% | 6.0 | Engineering and platform responsibilities are described; comparable fees need a proposal. |
| Platform independence | 10% | 9.0 | Open-source implementation and customer control are core principles, alongside Nebari. |
| Stability and scale | 10% | 6.5 | Continuity of a specialist engineering team, with a recent leadership and product-model transition. |
| Independent evidence | 10% | 6.5 | Public technical work is inspectable; much of the commercial outcome evidence is supplier-published. |
| Overall | 100% | 7.7 | Weighted average |
Source: AI Services Companies editorial assessment of OpenTeams: founders, team, and principles, OpenTeams: PyTorch engineering contributions.
Where OpenTeams fits
AI infrastructure and product-engineering leaders are the natural buyers. OpenTeams sits in the specialist deep-technical consulting tier, with platform deployment and support alongside engineering. Compare the targeted model, hardware, baseline performance, code ownership, and maintenance obligation. Public engineering work demonstrates capability but does not predict the performance gain on another workload.
For the same work, compare TechnoLynx for inference optimization and production AI R&D; deepsense.ai for deep-learning systems and model optimization; Enthought for scientific Python engineering and technical computing.
What clients and leaders say
“AI should not be a subscription. It should be an asset you own.”
“Open source makes your AI stronger, more secure, and more yours.”
“torch.compile is still where a large part of our team's effort goes.”
What people may not like
A simple business chatbot may not require its lower-level engineering expertise. Nebari delivery and custom performance work are different purchases with different support obligations.
