I'm a passionate AI Developer & Backend Engineer at Asyst Resources LTD, specializing in Agentic AI, Multi-Agent Systems, and Scalable Microservices. I live for the intersection of clean architecture and intelligent automation. Currently, I am focused on engineering Autonomous Agents, RAG systems, and bridging the gap between raw LLM logic and production-ready applications.
๐ Founder & Lead at Ijwi ry'Ikirundi AI
I founded and currently lead Ijwi ry'Ikirundi AI, an open-source organization dedicated to digitizing the Kirundi language. We are building the foundational datasets and models for the future of Kirundi AI.
- Organization: github.com/Ijwi-ry-Ikirundi-AI
- Hugging Face: huggingface.co/Ijwi-ry-Ikirundi-AI
Completed a comprehensive 6-week, 5-framework intensive building Autonomous Systems โ from raw LLM API patterns to production-grade multi-agent architectures.
| Framework | Project & Highlights |
|---|---|
| Raw Python | Professionally You Chatbot Built framework-free agentic patterns (Rerun, Eval loops) and deployed to HuggingFace. |
| OpenAI SDK | Deep Research & SDR Agents Engineered autonomous agents for complex research and sales with safety guardrails. |
| CrewAI | Dev Swarm Agency Modeled a full software team (Lead, Dev, QA) running code in Docker containers. |
| LangGraph | State machine-based orchestration with checkpointing, conditional routing, and browser automation agents. |
| AutoGen | AgentChat communication, Core infrastructure, and distributed multi-agent runtimes. |
| MCP | Built custom MCP servers/clients, 44+ tools across 4 AI traders with knowledge graph memory and real-time Gradio dashboard. |
- ๐ง๐ฎ Kirundi Open Speech & Text Dataset โ The first large-scale, open-source speech and text dataset for Kirundi. Leading a community of contributors to build resources for ASR, TTS, and Machine Translation.
- ๐ฎ Kirundi Contribution App โ A gamified web platform for crowdsourcing language data. Features "Easy" and "Medium" levels to make contributing fun and accessible. Play Live Here!
- ๐ Dev Swarm: Autonomous Software Agency โ A multi-agent AI system where autonomous agents (Manager, Backend, Frontend, QA) collaborate to build software. Features Dockerized code execution for safety and hierarchical task delegation using CrewAI.
- ๐งฌ Enterprise RAG ChatBot โ A fully containerized production-ready, framework-free RAG microservice.
- ๐ Burundian Language Identifier โ Custom-trained NLP model to classify Kirundi, French, Swahili, and English. Live Demo Here!
- ๐ฑ SamandariApp โ Flutter-based productivity hub powered by AI and customizable widgets.
- ๐ฐ E-Sama: Boutique & POS โ Production Flutter POS app with paying users. Supports 3 store types (Retail, Restaurant, Rental), 3 languages, offline-first with AES-256 encryption , subscription-based SaaS model.
- ๐งฉ Multi-Agent Orchestration: CrewAI, LangGraph, AutoGen, OpenAI Agents SDK, MCP.
- ๐ง AI Engineering: RAG, Vector Databases (Qdrant/Milvus), Embeddings, and Fine-tuning.
- ๐ฃ๏ธ Audio & NLP: Speech Recognition (Whisper/Faster-Whisper), TTS, and NLTK/SpaCy pipelines.
- ๐ก๏ธ Production Patterns: Evaluation loops, Guardrails, Dockerized Code Execution.
- ๐ Security: Cybersecurity, SOC Analysis (MITRE ATT&CK), Ethical Hacking (Kali Linux daily driver).
- โ๏ธ Backend: NestJS, Python (FastAPI/Django), and Microservices.
- ๐ฑ Mobile: Flutter (Hive, Provider, AES-256 encryption, Bluetooth ESC/POS printing).
- Rapid tech learner & adaptive problem solver
- REST API design & implementation
- Micro-services architecture (end-to-end)
- Clean code, documentation & testing
- Security-first mindset (Ethical Hacker at heart ๐)
๐ง๐ฎ The first large-scale, open-source speech and text dataset for Kirundi language. A community-driven initiative building AI capabilities for 12M+ Kirundi speakers through collaborative data collection, featuring ASR, TTS, and machine translation capabilities with comprehensive documentation and automation tools.
๐ฐ Production Flutter POS & shop management app with paying users. Supports 3 store types (Retail, Restaurant, Rental), 3 languages (EN/FR/SW), offline-first architecture with AES-256 encryption, Bluetooth thermal printing, CRM & loyalty, and subscription-based SaaS monetization.
A comprehensive backend solution micro-service for user authentication, authorization, and management using NestJS with Keycloak, Vault, Swagger, and more. Offers robust RESTful APIs with security best practices, role management, and seamless integration for scalable applications.
| Degree / Diploma | Institution | Year |
|---|---|---|
| B.S. Software Engineering | Lake Tanganyika University | 2021 - 2024 |
| Diploma in Nursing | Ecole Paramedicale de Gitega | 2017 - 2021 |
| Course / Bootcamp | Key Skills Acquired |
|---|---|
| AI Engineer Bootcamp 2025 | NLP & Speech: Tokenization, Lemmatization, Whisper implementation. Vector Search: RAG pipelines with Pinecone & Faiss. Prototyping: Built AI Interviewer with Streamlit & Session State. Ethics: Bias mitigation and Responsible AI lifecycle. |
| Agentic AI Engineering | Systems: Multi-agent orchestration, Tool calling, Evaluation loops. Frameworks: CrewAI, OpenAI Agents SDK. |
- โจ๏ธ You canโt beat me at typing
- ๐ฅ Former Nurse turned Engineer โ I treat production bugs like medical emergencies.
- ๐ต Music producer & late-night coder
- ๐ญ Sweet tooth certified โ I debug with chocolate
- ๐น๏ธ Gamer, hacker, and tech philosopher
