Redko Labs research notes · October 13, 2026
Why this field deserves attention
Berkeley’s Sky Computing Lab studies data-intensive systems. Stanford’s shared nanofabrication and characterization facilities support hardware and materials research. Software and evaluation tools can offer an earlier starting point than manufacturing new chips.
One company idea to explore
A hardware-aware inference profiler for one class of models that shows the latency, memory, energy, and quality trade-offs of deployment choices.
A possible first customer
AI product teams with a real inference bill, or engineering groups deploying models on constrained devices.
The first experiment
Benchmark one workload across a controlled set of configurations. Publish reproducible measurements, include quality regressions, and show savings on a customer’s permitted workload.
What a founding team might bring
Computer architecture, compilers, distributed systems, electrical engineering, optimization, and machine learning systems.
What must be resolved
Measure end-to-end cost and quality, not only throughput. Chip design and fabrication need a separate capital plan, specialist tooling, rights, and facilities access.
This concept is a starting hypothesis from Redko Labs. Bring a different idea, challenge the assumptions, or apply without an idea. The work begins with evidence.
Bring the question.
Build the evidence.
Apply on a rolling basis, with an idea, a company, or the ambition to find your starting point.
Sources and research communities
Universities and organizations are independent of Redko Labs. Their inclusion describes public research activity, not a partnership, endorsement, facility entitlement, or recruitment agreement.
See how the curriculum supports this work