We’re building autonomous AI systems that operate carrier-scale networks.
A phone won’t register. You upload a PCAP and QXDM trace — NeuralTap decodes the signaling, finds the Registration Reject cause (PLMN not allowed), and tells you exactly which subscriber profile field to fix.
Handovers keep failing on Sector 2. NeuralTap correlates PM counters, UE traces, and cell config — finds a missing neighbor relation and gives you the exact config line to add.
A Samsung vDU is losing sync. NeuralTap traces it from the PTP sidecar OOMKilling to DPDK hugepage exhaustion and shows you the resource limit fix.
PDU sessions are failing on a sector. NeuralTap follows the signaling through the SMF, checks the N3 transport path, and finds the VLAN tag mismatch on the switch.
During a segment routing cutover at a new agg site, NeuralTap snapshots pre-migration state, validates IS-IS, BGP-LU, and EVPN end-to-end, and flags every missing route and stale peering.
Cepstrum Labs is the organization behind NeuralTap — applied AI researchers and engineers from Stanford, Berkeley, and Purdue, working alongside people who have spent 30+ years solving the hardest technical problems inside tier-1 carriers.