Back to About

Yussuf Hersi

AI Architect & Senior Software Engineer
Nairobi, Kenya • yussuf@hersi.devhersi.dev
GitHubX (Twitter)


Profile

Senior Software Architect transitioning into AI Engineering (2026 Phase). Expert in building agentic orchestration loops, hybrid retrieval systems (GraphRAG), and production-grade AI platforms. Proven track record of architecting scalable distributed systems and shipping high-impact SaaS products.

Core Expertise

  • AI Engineering: Agentic Design Patterns, Multi-Agent Orchestration (CrewAI, LangChain), Long-Context Window Management.
  • Retrieval Architecture: Hybrid GraphRAG (Neo4j + Vector), Semantic Caching, Reranking Pipelines.
  • Backend Architecture: Distributed Systems, Microservices (Go, Rust, Python), Scalable APIs.
  • Intelligence Ops: Tracing & Observability for non-deterministic systems, LLM Evaluation Frameworks.

Professional Experience

Founder & AI Lead | ContextBytes

Jan 2025 — Present

  • Architected a production-grade AI Cognitive Architect for processing long-form media (PDFs, Videos, Articles).
  • Implemented a hybrid GraphRAG system using Neo4j and Vertex AI, improving retrieval accuracy by 40% compared to vanilla vector search.
  • Developed autonomous agent loops for automated knowledge synthesis and memory management.

Founder | Pesastream

2024 — 2025

  • Built a fintech SaaS platform simplifying African payments via Mpesa integrations.
  • scaled the platform to handle mission-critical financial transactions with 99.9% uptime.

Software Engineer | Mursal Technologies

2023 — 2024

  • Designed and maintained scalable backend services using Go and PostgreSQL.
  • Optimized internal developer tools, reducing CI/CD latency by 30%.

Key Projects

  • Aftahan NLP: A specialized linguistic engine for Somali language processing, enabling deep localization in LLMs.
  • NextDeploy: A vendor-independent deployment engine for Next.js applications using Go and Docker.

Technical Stack

  • Languages: Go, Rust, Python, TypeScript.
  • AI/ML: OpenAI API, Vertex AI, LlamaIndex, LangChain, Neo4j.
  • Infrastructure: Docker, Kubernetes, PostgreSQL, Redis, AWS/GCP.

Tip: You can edit the source at src/content/cv.md