AI infrastructure that runs on your terms.
Independent systems lab building tools for code understanding, vector search, file verification, and agent infrastructure — designed for privacy, portability, and self-hosting.
Three tools in alpha. Free to use.
Production-grade infrastructure you can run on your own hardware. No accounts, no pricing tiers, no lock-in.
Chunker
Upload a codebase, get back a searchable knowledge graph. Semantic chunking that understands your code's structure — not just splitting by line count.
qql-go
A single command to query, inspect, and manage your Qdrant vector database. No SDK code needed. Works from the terminal, CI pipelines, or shell scripts.
Stamp
Prove a file existed at a point in time — without ever uploading it. Cryptographic receipts and public transparency logs you can verify independently.
Four more on the way.
Being built in the open. No release dates, no waitlists — they'll ship when they're ready.
repo-cli
Give AI agents a safe, structured way to understand and modify code. 5 autonomy modes — from read-only exploration to full patch workflows with test validation.
Mono-Auth
Self-hosted authentication for humans and machines. Single sign-on, API keys, and machine-to-machine tokens — all running on your own server.
GCM
Let AI agents write and run code safely. Every action is analyzed before execution, and every tool call passes through a policy gate — no exceptions.
VIA
Catch anomalies in your systems before they become incidents. Real-time detection that learns from your data and gets smarter with every signal.
"Most tools ask you to trust someone else's computer. We don't."
Trust this auth provider. Trust this cloud database. Trust this agent platform. Trust the supply chain.
Veristamp builds tools you can inspect, self-host, and verify — so you never have to take anyone's word for it.
Your data stays yours
Files, keys, and sensitive data never leave your machine unless you choose to share them. No telemetry. No tracking.
Runs on your hardware
Every tool works on a VPS, a Raspberry Pi, or your laptop. No mandatory cloud. No vendor lock-in.
Verify, don't trust
Cryptographic receipts, signed proofs, append-only logs. You can check the math yourself — no need to believe us.
Lightweight and affordable
Optimized to run on minimal hardware. No bloated frameworks. No infrastructure that costs more than the problem it solves.
Open formats, no traps
Exportable data, documented APIs, standard formats. Take your data and leave anytime. No proprietary prisons.
Build logs, architecture notes, and lessons from the field.
Deep dives into AI agents, RAG pipelines, identity fabrics, self-hosted infra, and the tooling we ship.
Governed Code Mode: From Tool-Calling to Zero-Trust Execution
How Cloudflare, Anthropic, and Google are converging on a new pattern — and what a governed variant looks like.
Qdrant & Toolingqql-go: A Single-Binary CLI for Qdrant Operations
Query, inspect, and manage Qdrant from the terminal — no SDK code needed.
RAG & KnowledgeSchema-First Extraction for LLM Wikis with GLiNER2
Turn messy RAG pipelines into persistent, queryable knowledge graphs.
Srimon Danguria
Independent systems builder. Qdrant Star. Building agent infrastructure from first principles.
Questions, feedback, or just curious?
Whether you're trying out Chunker, debugging a qql-go query, or want to talk about agent infrastructure — I read every email.
Send an email