Data-sovereign AI architecture · UK · US · UAE
AI that runs where your data lives.
I architect on-device, offline-first, and air-gapped AI systems for teams that can’t — or won’t — send sensitive data to someone else’s cloud. Nineteen years designing mobile and cloud platforms that scale to millions.
DPIIT-recognised LLP · Outside-IR35 ready · CSPO · PRINCE2 · MBA
The problem with most AI consulting
Most AI advice assumes one thing: your data leaves your building and lands in someone else’s cloud. For healthcare, defence, finance, and privacy-first products, that assumption is a dealbreaker — regulatory, commercial, or both.
I build the other path. Models that run on the device, offline, and under your control — no data egress, no third-party inference, no dependency on connectivity you don’t have. When the cloud genuinely is the right answer, I’ll tell you that too. But sovereignty is the default, not the afterthought.
Sovereign by default, cloud when it earns it
On-device & edge AI
Quantized LLMs that run on phones and edge hardware — Gemma-class models, QLoRA fine-tuning, swappable LoRA language adapters. Inference at the edge, not in a datacentre.
Offline-first & data-sovereign architecture
Air-gapped and offline-capable systems with zero data egress. Your data never leaves your control. Built for regulated and privacy-critical domains.
Mobile architecture at scale
Nineteen years of React Native and native mobile architecture — the platform layer behind apps serving hundreds of millions of users.
Cloud & GenAI systems (when cloud is right)
Production RAG, multi-agent systems, serverless and multi-region cloud — designed to hold their shape under load.
Fractional AI architect / CTO advisory
Senior architecture leadership on a contract basis — stack decisions, architecture reviews, roadmaps. Outside-IR35 for UK engagements.
What sovereign AI looks like in production
Offline clinical AI for frontline health workers
An on-device clinical decision-support system running fully offline on low-end Android hardware, with multilingual on-device language support. Designed for connectivity-poor settings under a medical-device compliance framing. No cloud, no data leaving the device.
Embedded sovereign data & search layer
A sovereign embedded database and on-device vector search engine — retrieval and semantic search that run entirely on-device, with no cloud dependency. Shipped as production components (published open-source packages).
Edge AI SDK
A production edge-inference SDK: C++ core with a mobile bridge layer, enabling on-device vector storage and retrieval inside React Native apps.
On-device LLM in a consumer app
A consumer discipline-and-practice product with a Gemma-class language model running entirely on-device — private by construction, functional without a network.
Scale platforms (under NDA)
Architecture and delivery for a major OTT/streaming platform, a leading e-commerce platform, and a ride-hailing platform — systems serving hundreds of millions of users.
UK proptech
Technical architecture and documentation for a UK proptech venture building autonomous property operations.
- 19+ years architecting mobile, cloud & AI systems
- Past roles incl. JLL, Rackspace, Robosoft, Accion Labs
- CSPO · CSM · PRINCE2 · MBA (Leadership & Project Management)
- DPIIT-recognised, GST-registered Indian LLP — invoicing-ready for international clients
- Outside-IR35 engagements available for UK clients
How to engage
Architecture sprint
Fixed-scope engagement: audit, design, and a decision-ready architecture for your AI/mobile system.
Fractional AI architect
Ongoing retainer; senior architecture on tap without a full-time hire.
Outside-IR35 contract
UK contract engagements, compliant and ready to start.
Design + build pod
Architecture through to delivery with a scrum-pod team, scaled to the engagement.
About
Satyam Kumar Das — Founder & Solutions Architect, AppScale LLP
Nineteen years building mobile, cloud, and AI systems — from the platform layer behind apps at massive scale to on-device AI that runs with no cloud at all. My architectural philosophy is simple and consistent: offline-first, on-device, data-sovereign. AppScale LLP is a DPIIT-recognised product studio and AI consultancy working with founders and enterprises across the UK, US, UAE, and APAC.
Have a system that can’t go to the cloud?
That’s exactly the kind of problem I like. Tell me what you’re building and where the constraints are.