Services
Node/TypeScript APIs, event-driven workflows, and distributed backends
Node/TypeScript APIs, event-driven workflows, and distributed backends
Schema design, caching layers, and data integrity at scale
AWS, containers, observability, and global delivery infrastructure
Composable UIs, design systems, and micro-frontend surfaces
Eight years of backend architecture, APIs, and production reliability. Sole architect of HustleSasa backend services scaling 1M+ payment transactions daily at 99%+ uptime. Node/TypeScript microservices, payment rails, and cloud-native infrastructure serve high-traffic platforms across African markets. Live entertainment and online ticketing workloads are a focus area. Software factory work covers agent harnesses, validation contracts, and eval gates that help teams ship verified changes.
Node/TypeScript APIs, payment services, and event-driven backends on Postgres, Mongo, Redis, and AWS/Kubernetes. Vue, React, and Svelte surfaces plus agent harnesses and validation gates when teams adopt agent-native delivery. Systems scale across every layer — platform first.
Node/TypeScript services with clear domain boundaries. REST and GraphQL APIs, resilient messaging, and observable pipelines stay reliable under load.
Payment rails, rate limiting, caching, and uptime discipline for high-traffic platforms — the same patterns that keep ticketing and live-event surges healthy.
Event-driven architectures on AWS with Docker/Kubernetes delivery — API gateways, background workers, and infra that scales without idle waste.
Agent harnesses, validation contracts, and orchestration loops. Intake, implementation, checks, and review gates help teams produce verified changes without linear headcount.
Module federation and independent frontend teams shipping Vue, React, and Svelte apps that compose into a unified product surface.
Full-stack hospital management system for clinics in Ghana: patient registration, OPD triage, billing, and role-based access
High-performance automotive data autofill engine rewritten in Rust
Agent workflow: Clay enrichment and signal triggers feed n8n orchestration and HubSpot CRM. LLM agents handle lead research, routing, and outbound. Eval gates run before any prospect contact.
Serverless platform for automating approval and workflow routing across departments
Visit ↗GraphQL library for building schema-first APIs with minimal boilerplate
Repository ↗GitHub Action for deploying SST applications to AWS with zero config
Repository ↗TypeScript is home base. Other languages fit when the problem demands a different shape: memory safety, numeric performance, or AI-native tooling.
Primary language for full-stack delivery: typed Vue/React frontends, Node services, shared contracts, and most production systems.
Systems programming experimentation: memory safety, concurrency, and performance patterns before production use.
AI agent pipelines, harness scripts, data tooling, and backend prototyping — the default reach for LLM-powered systems.
Early exploration for machine learning workloads: Python-like ergonomics with a path toward bare-metal performance for model-centric systems.
Software factory engineering designs the system that produces software. Intake rules, agent orchestration, validation harnesses, and review gates turn backlog into verified output. Platform discipline keeps the line reliable under load.
Orchestrator, worker, and validator roles with serial gates. Tasks carry acceptance criteria before any agent edits code.
Numbered checks gate done from proven. Eval gates run before merge, release, or outbound contact.
File-backed state and loop-until-goal runs survive crashes and long jobs without chat memory.
CI/CD pipelines, sandboxes, observability, and autonomy policies control what agents can change and ship.
Production AI systems sit on solid platform foundations: agent orchestration, evaluation frameworks, inference pipelines, and reliability layers that turn demos into production. LLM eval and QA work applies the same gate pattern as software factory validation contracts.
Multi-step LLM workflows for research, enrichment, routing, and follow-up. Tool use, guardrails, and human handoff when confidence is low.
Eval frameworks for LLM outputs: quality scoring, regression detection, and production safety gates before release.
Latency, context windows, batching, and deployment patterns for fast, observable, dependable models in real products.
Transformers, attention, embeddings, and architecture choices that shape capability and cost. Production systems are designed with limits in mind.
Engineering is the craft, but balance keeps the edge sharp. Competitive games, pitch time, and the occasional jog keep me grounded outside the terminal.
Fast-paced BR: movement mechanics and team coordination.
Football on the couch: tactics, seasons, and competitive matches.
Tactical FPS: map knowledge, operator synergy, and clutch rounds.
Laps for clarity: low-impact cardio and a break from the screen.
Pickup matches with friends on grass.
Casual runs and shoot-arounds: competition without league fees.
Morning or evening runs for focus and fitness.
Open to Lead Back-End, platform engineering, and software factory roles. AI systems consulting remains available.