Software engineer specialising in LLMs, RAG systems, and production AI — building for Uganda and beyond.
CORE TECHNOLOGIES
hover planets to explore
I build AI systems that run in production, backed by nearly a decade of software engineering experience. Rather than running local experiments, I focus on integrating LLMs and RAG systems into scalable infrastructure
Commerce Engine is the output: an AI-powered commerce infrastructure that seamlessly orchestrates orders across WhatsApp and webchat for Ugandan SMEs without requiring extra staff.
I also maintain a technical blog where I write about AI engineering and LLM development. The blog features a full RAG implementation — you can chat with my posts to explore concepts like semantic search, vector databases, and prompt engineering in practice.
“I build things that run.”
Interactive AI Architecture Lab
Live, production-grade micro-demos illustrating core LLM and NLP engineering paradigms. Open a card to interact with real language models in real time.
BERT vs GPT
NLPBidirectional vs Unidirectional Attention
Click any word in a sentence. The AI explains what context BERT sees (both directions) vs what GPT has at that position (left only).
Prompt Engineering
PromptingZero-shot · Few-shot · Chain-of-thought
Enter any classification task. Three prompting techniques run against the same input — compare the actual prompts sent and the responses received in real time.
RAG Pipeline
RAGIngest · Retrieve · Generate
Paste a document, watch it chunked and indexed, then ask a question. The system retrieves the most relevant chunks and generates a grounded answer.
Token & Cost Management
ProductionCount · Compare · Optimise
Type any prompt to see live token estimates and cost across 5 models. Use the AI optimizer to trim tokens without losing intent.
Project Commerce Engine
AI-powered commerce infrastructure for Uganda's SMEs — orders over WhatsApp, webchat, and voice without hiring more staff.
Customers order in natural language over WhatsApp or webchat. The AI understands intent, builds the cart, and confirms the deal without human involvement.
Customer Ordering
Customers order in natural language over WhatsApp or webchat. The AI understands intent, builds the cart, and confirms the deal without human involvement.
Smart Import
Upload a menu image, PDF, or URL — AI builds your full product catalog with business rules in under 5 minutes.
Merchant Dashboard
Product Studio gives merchants full control over catalog health, product templates, preconfigured variants, and business rules — no code required.
HITL Command Centre
Flagged conversations appear in real time. Merchant takes over, edits the draft order, hands back to the AI in one flow.
Tech stack
Customer: "I want drinks from jj foods" — AI surfaces product images from the live catalog instantly
AI/ML applied in this project
Document Intelligence
Vision model parsing of PDFs and menu images to extract structured catalog data
Intent Classification
NLP to map free-text customer messages to catalog items and commerce actions
Structured Output
LLM constrained to emit valid order JSON matching the deterministic commerce schema
Multi-turn State
Conversation context managed across async WhatsApp and webchat message sequences
Early access · September 2026
Be among the first merchants on Commerce Engine.
The platform launches for Ugandan merchants in September 2026. Join the waitlist now and lock in founding member pricing — a significant discount before public launch.
Contact
Let's build
production AI.
AI/ML engineering
LLM integration, RAG systems, production deployment
Consulting
Architecture reviews, AI strategy, prompt engineering
African market focus
Products built for emerging market infrastructure
Full-time roles
Engineering positions in AI-first teams
Request resume
Share your email and I'll send it to you directly.