How we think.
What we've solved.
Real engineering problems with real solutions. We publish what we learn â technical depth, not hype.
How we build software
ð
Research before code
Every product begins with deep study â reading RFCs and specifications, analyzing existing solutions, understanding where they fall short.
âïļ
Build from specifications
RFC 6749, RFC 7519, SAML 2.0 Core, SCIM RFC 7644, SMART on FHIR. No wrappers. Full control over correctness.
âĄ
Ship weekly, break nothing
Zero-downtime deployments, comprehensive testing, structured audit logging. Regulated workloads demand reliability.
ð§
AI is the architecture
Threat detection, semantic analysis, clinical intelligence. Not a marketing label. The actual engineering.
ðŽ
Go deep when it matters
eBPF for kernel auth at ~45Ξs. We don't reach for abstractions when the problem demands precision.
ð
Open standards
We build on open protocols. Customers own their data, identity, and infrastructure. No lock-in.
Problems we've solved
Stateless JWT validation in eBPF bytecode
RS256/ES256 JWT verification compiled into eBPF programs at the socket layer. Per-packet auth at kernel level, ~45Ξs. No sidecars.
Cross-cloud credential exchange via OIDC
Single auth event â short-lived credentials for AWS STS, GCP Workload Identity, Azure, OCI. Zero static keys for humans and machines.
Behavioral threat detection across identity layers
Correlating app login, cloud IAM, service calls, and network signals. Impossible travel, credential stuffing, lateral movement.
Multi-tenant identity with per-tenant branding
Isolated OIDC config, custom branding, independent modules, separate audit streams â on shared infrastructure.
Full SAML 2.0 SP + IdP from specification
Assertion generation, signature verification, attribute mapping, relay state. SP-initiated and IdP-initiated flows from OASIS spec.
Semantic analysis beyond AST parsing
Code understanding that reasons about intent, cross-file relationships, and architectural patterns â not just syntax trees.
Exploring
Natural language â eBPF policy
"Only payments can reach billing DB" â compiled to kernel bytecode.
Continuous identity verification
Behavioral signals that verify users throughout sessions â not just login.
LLM-powered audit queries
"Who accessed prod credentials this week?" â from structured event data.
Semantic code understanding
Models that reason about intent across entire repositories.
AI clinical interoperability
Language models bridging healthcare data standards.
Technology
By the numbers
How we use AI to ship fast
If I had an hour to solve a problem, I'd spend 55 minutes thinking about the problem and 5 minutes thinking about solutions.
â Albert Einstein