Practical systems for AI, identity, automation, and digital trust.
Veristamp is an independent systems lab building tools for authentication, agent workflows, knowledge infrastructure, verification, and automation — designed with a bias toward privacy, control, and low-cost infrastructure.
M2M auth fabric & 0-RTT revocation filters.
BuildingPrivacy-first digital receipts & timestamp logs.
AlphaControl plane & logs for governed agent swarms.
BuildingStructured local memory and retrieval pipelines.
InternalActive Systems
We build independent tools mapping to key infrastructure concerns. Here is the active inventory with their public development status.
Mono-Auth
Self-hosted OIDC identity control plane with 16-bit Cuckoo filters for 0-RTT revocation checks.
Veristamp Stamp
Privacy-first digital receipts, integrity verification, and epoch-bounded Merkle log archives.
AgentHub
Non-linear collaboration engine and Content-Addressed Storage (CAS) for autonomous AI swarms.
Veristamp Chunker
AST-guided semantic markdown and code chunking visualizer powered by tree-sitter.
VIA (Incident Atlas)
Multi-tier real-time anomaly detection engine and adaptive Thompson Sampling agent supervisor.
Knowledge Base
Dual-graph knowledge base integrating Postgres structural trees and Qdrant semantic similarity.
Modern software is held together by too many fragile promises.
We are told to trust: trust this login broker, trust this audit log, trust this third-party script, trust this black-box AI model, trust this cloud pipeline.
Veristamp is built around a different idea: important workflows should be inspectable, automatable, privacy-aware, and cheap enough for independent builders to run.
Core Principles
These engineering constraints bind the entire suite. If a tool violates these values, it gets refactored or deleted.
Small, sharp systems
Each tool should solve a real workflow clearly instead of becoming a bloated platform.
Privacy before convenience
Sensitive data should stay out of the system unless there is a strong reason to collect it.
Proof over promises
Important actions should leave inspectable, verifiable records.
Cheap to operate
Useful infrastructure should be able to run on lean, efficient systems before scaling up.
Portable by default
Prefer clear APIs, exportable data, open formats, and low lock-in.
From the Blog
Engineering notes, architecture breakdowns, build logs, and practical lessons from building Veristamp systems.
View all postsThe Identity Trap: Why Building Custom IdPs is Risky
How mono-auth implements a micro-budget identity fabric and OIDC-aligned O-RTT revocation filters without bloat.
AI SystemsGoverned Code Mode: AST-Guided AI Safety
Designing lightweight AST-grepping lint gates to sandbox autonomous agent tool executions and prevent secret leaks.
Built by Srimon Danguria
Independent systems architect working across distributed AI operating systems, identity fabrics, local knowledge infrastructure, and security verification tools. Drawing from a background in Electronics and VLSI microelectronics research, I design resilient, low-overhead digital systems from first principles.