◉ SANDHAUS.SYS
--:--:--

> NEURAL UPLINK _

BOJAN SANDHAUS

SAAS, IT & AI CONSULTANT · ENTERPRISE SYSTEMS, CALM DELIVERY

SIGNALSTRONG
MODELINITIALIZING…
LOCAL TIME--:--:--
STATUS● ACCEPTING MISSIONS
SCROLL TO DECRYPT
01

HUMAN IN THE LOOP

IDENTITY // WHOAMI

> whoami

I’m a multilingual SaaS, IT and AI consultant with a passion for turning complex enterprise challenges into seamless, scalable digital solutions — and AI hype into systems that actually ship.

With over 7 years of experience in software implementation, agile project delivery, and organizational change, I help companies align teams, improve user adoption, and unlock real business value.

Whether working with Fortune 500s or startups, I bring a calm, clear, and collaborative approach — always focused on results, efficiency, and lasting impact. Lately that means helping teams adopt AI where it genuinely pays off, and skip it where it doesn’t.

Portrait of Bojan Sandhaus
▸ OPERATOR: B. SANDHAUS
CLEARANCE: ENTERPRISE
0+YEARS SHIPPING
500FORTUNE-LEVEL CLIENTS
100%CALM UNDER PRESSURE
15SYSTEMS OPERATED
02

CAPABILITIES

MODULES // LOADED
03

THE STACK

DEPLOYED & OPERATED // PRODUCTION-TESTED

Systems I deploy, run and maintain in production, for my own lab and for clients. If it's on this list, I can implement it for you.

⌁ Hermes Agent AI AGENTS Autonomous agent runtime that plans, uses tools and delegates to subagents across real business workflows, from ops to outreach. AGENTS ON DUTY ◈ Deepseek Harness AGENTIC CODING Open-source agent harness where everything is a plugin. An MIT-licensed rival to Claude Code: repo editing, shell, subagents and model freedom. DAILY DRIVER ◔ Hindsight AI MEMORY Continual-learning memory that reasons over interactions and returns only the context that matters. Agents that remember without drowning in tokens. NEVER FORGETS ◉ Jev Decisions AGENT OVERSIGHT My open-source plugin that makes agents check plans, verify evidence and flag risky actions before they run. A decision journal for every call. EVERY ACTION JUDGED ◍ Ollama LOCAL LLM Open-weight models on your own hardware. Private inference, no API meter, no data leaving the building. MLX on Apple Silicon, vLLM when a whole team hits it. RUNNING ON SILICON ⟁ n8n AUTOMATION Workflow automation connecting apps, APIs and AI agents. The glue between systems that were never meant to talk. WORKFLOWS HUMMING ⬡ Bitwarden SECURITY Password management plus Secrets Manager, credentials and API keys under control across the whole team. VAULT SEALED ▨ Paperless-ngx DOCUMENTS Document management that scans, OCRs and indexes paper into a searchable archive. ZERO PAPER ≣ Hister SEARCH & RETRIEVAL Personal search index over your own data, the retrieval backbone that makes company knowledge actually findable. EVERYTHING INDEXED ⌕ SearXNG SEARCH One query fans out across dozens of engines at once, the whole searchable web in a single result page. EVERY ENGINE AT ONCE ⚑ Karakeep KNOWLEDGE Bookmark and content archiving, building a knowledge base your team can actually search. NOTHING LOST ≋ FreshRSS MONITORING RSS aggregation for market and industry monitoring, self-curated feeds instead of an algorithm's picks. FEEDS FLOWING ▞ BentoPDF DOCUMENTS Client-side PDF toolkit: merge, split, sign and redact without documents leaving the browser. DOCS STAY LOCAL ⇄ VERT FILES Blazing-fast file conversion straight in the browser. Any format to any format, no queues, no waiting. CONVERTED IN SECONDS ⌂ Home Assistant THE LAB AT HOME The one that runs my house. Where I break things first, so client systems stay boring. PROVING GROUND
04

SIGNAL QUERIES

FAQ // ANSWERED
Q.01 What does an AI consultant actually do? +
> RESPONSE

Cuts through the hype, finds the two or three places AI genuinely pays off in your business, and ships them: agents, automation, retrieval over your own documents. Then trains your team so it sticks after the engagement ends.

Q.02 Can AI run on our own hardware instead of the cloud? +
> RESPONSE

Yes. Open-weight models have reached the point where a single workstation or a Mac Studio cluster runs serious workloads. I deploy Ollama, MLX or vLLM on your hardware, wire in retrieval over your documents, and your data never leaves the building.

Q.03 What does a local LLM rollout cost? +
> RESPONSE

Far less than a year of enterprise API seats for most teams. Hardware is a one-time cost, the models are open-weight, and there is no per-token meter running. A typical pilot is measured in weeks, not quarters.

Q.04 How do you stop AI from hallucinating in business answers? +
> RESPONSE

Never let the model answer from memory. It answers from your verified data through retrieval, cites its sources, and an oversight layer checks plans and evidence before anything acts. Answers you can audit, not vibes.

Q.05 Do you work with companies outside the Netherlands? +
> RESPONSE

Yes. I am based in Utrecht and work across Europe and remotely worldwide, in English, Dutch and German. Multilingual delivery is a core part of the practice.

05

CONTACT

CHANNEL // OPEN

> open_channel --secure

LET’S BUILD SOMETHING
THAT WORKS.

↗ LINKEDIN