In development
KMS / Ramp
Ramp is a developer onboarding tool designed to help engineers understand unfamiliar codebases. It builds bounded, evidence-backed views of how implementation surfaces relate, then uses those candidates to generate structured onboarding plans. The system is designed to distinguish observed code relationships from assumptions that still require human verification.
Engineering highlights
- Company-scoped code indexing and evidence extraction
- Evidence composition across connected implementation surfaces
- Deterministic candidate selection and onboarding plan generation
- Persistent onboarding plans and progress
Implementation notes
Evidence-grounded code understanding
An early approach to codebase understanding relied too heavily on directory proximity and path naming. Those signals were insufficient to establish that files participated in the same engineering workflow. Ramp moved toward using indexed implementation surfaces and available code relationships to construct bounded workflow candidates.
The evidence model separates direct relationships from derived and inferred ones. A relationship is not automatically treated as proof of runtime behavior. The system preserves uncertainty so that engineers can inspect the underlying source, verify control and data flow, and identify gaps rather than receiving an apparently complete but unsupported explanation.
Deterministic onboarding plans
Workflow candidates are evaluated and sequenced through deterministic logic. Eligibility, risk, prerequisites, and verification signals are kept distinct from language-model output. This separation is intended to make plan generation more inspectable and reduce the risk of an LLM presenting unsupported relationships as established facts.
Generated onboarding plans and their progress are persisted so that the work can be revisited. The current implementation represents an evolving product direction; it has not established measured improvements in onboarding time, developer productivity, or production-scale reliability.
Technical areas
Python · Go · Supabase · LLM integration