Data residency, by design
Sensitive banking, health, and public-sector data never leaves its legal borders. Models run on our local cluster or directly on-premise at the client.
Deep tech · Applied AI research · Sovereign infrastructure
DipR AI is an applied research lab natively building AI for African communities so no external entity can ever pull the plug.
Why we exist
African entities are forced to rely on closed-source, foreign-hosted AI black boxes. This breaks data residency laws, exposes sensitive metadata to extraterritorial reach, and primarily bills local businesses in foreign currencies.
Sensitive banking, health, and public-sector data never leaves its legal borders. Models run on our local cluster or directly on-premise at the client.
A foreign API can change pricing, alter filters, or pull access overnight. Open-weight models owned and adapted locally make that impossible.
Flat-capacity enterprise tiers billed in XOF decouple West African businesses from fluctuating dollar-denominated token markets.
“Small models for Africa” is a mathematically flawed compromise. We train at the parameter scales dictated by empirical scaling laws — then distill for frugal inference.
The platform
From physical hardware to model weights, we own the full AI value chain — the same structural logic as elite foundational labs, executed with infrastructure realism.
Infrastructure layer
A proprietary, high-density NVIDIA cluster collocated in a regional Tier-III facility in Grand-Bassam’s VITIB SEZ. Uncompromised data residency, immunity from foreign regulatory interference, and absolute control over the regional AI value chain.
Research layer
No API wrappers. Deep-tech research on model compression, architectural efficiency, and advanced quantization — turning raw compute into high-performance base models built for data-scarce, infrastructure-constrained environments.
Enterprise layer
Sovereign infrastructure and research power three high-impact pillars for immediate market adoption across regulated enterprises, governments, and the regional mid-market.
Open research
The open-weight, trust-by-control playbook: we release models and publish at elite venues to establish Francophone West and Central Africa as a hub for the next generation of AI — and to attract the continent’s best researchers home.
Speech processing · RL optimization · Efficiency
Audio-native tokenization frameworks trained directly on regional data — Baoulé, Dyula, and beyond — instead of passing African languages through a French or English textual filter first.
ACL · EACL · NAACL
Model compression · Hardware-aware NAS · Edge inference
Open-sourced algorithms producing highly portable models without sacrificing performance — high-fidelity AI on standard consumer devices over fluctuating edge networks.
ICLR · ICML · NeurIPS workshops
Agentic IT auditing · Data privacy · Alignment
Peer-reviewed case studies on sovereign data structures, plus an open-source agentic auditing sandbox for highly regulated B2B and GovTech clients.
ACM AIES · ACM EAAMO
Monitoring · Early-warning systems
Contributions to open-source initiatives such as InstaGeo, TAHMO, and AfriClimate AI, turning regional operational constraints into recognized scientific contributions.
Open-source & domain venues
The team
Alumni of world-class global and African universities, with advanced postgraduate and industry experience in AI, digital trust, and cybersecurity. We offer what elite talent actually wants: compute, intellectual liberty, and hard problems — with international-grade vesting equity.
CEO & Co-founder
AI engineer; incoming PhD student in AI Safety (INP-HB); former CMU-Africa research associate; lead writer of a $250k grant-winning proposal on African mother tongues; AI analyst at IFC.
Chief Scientist & Co-founder
Applied AI research leadership across global and African institutions.
CTO & Co-founder
Senior Software Eng, Cloud Solution Architect, specialized in the design and build of distributed systems; ESATIC alumnus.
Research
PhD in AI — vehicular edge computing, task offloading, deep reinforcement learning, trust management, federated learning.
Research
PhD in Free Space Optical communications.
We are raising a $500k pre-seed round complemented by NVIDIA Inception compute, on the path to a $10M seed. Investors, Tier-1 enterprises, researchers — let’s talk.