Prince Kofi Boateng.
Full-Stack Software Engineer.

Online pharmacy marketplace with a full delivery-dispatch network.

A production e-pharmacy platform where customers order medication and verified pharmacies fulfil it, backed by a purpose-built dispatch layer that routes deliveries to independent rider fleets. I built the customer storefront, the pharmacy back office, and the dispatch/rider portals — each with its own auth, wallet and real-time delivery state.

• Multi-portal system: customer, pharmacy, dispatch-company admin and rider apps
• Wallet + transaction ledger and per-fleet financial reconciliation
• Delivery lifecycle: claim, assign rider, live status updates
• Dockerised deployment with Caddy handling automatic TLS

An offline-capable point-of-sale and pharmacy management system.

A point-of-sale and inventory platform for pharmacies that keeps working when the internet doesn't. Built as an installable PWA so counter staff can ring up sales, manage stock and generate reports offline, syncing when back online. Currently being rebuilt on Nuxt 4 with a fully typed Pinia state layer.

• Installable, offline-first PWA for unreliable-connectivity environments
• Typed domain layer with Pinia stores and vee-validate forms
• Inventory, sales, and reporting workflows for retail pharmacy
• Migration-driven rebuild with typecheck + test gates

Connecting African students and SMEs through verified digital resumes.

A career platform that lets students build a verified digital resume, showcase skills and find short-term and part-time work, while giving SMEs access to affordable, vetted talent. Spans a web platform, a backend API and a companion mobile app.

• Verified digital resume and skills profiles for students
• Matching for short-term / part-time opportunities
• Employer side for SMEs to discover and hire talent
• Web platform + Node backend + Ionic/Vue mobile app

A multi-agent pipeline that plans, implements and reviews code with different AI models.

A personal working tool that runs AI-written code through a Plan, Implement, Review, Human Gate and Test pipeline. Each stage is routed to a different model (Claude, GPT, GLM, DeepSeek) by cost and capability, with review gates that check AI output for correctness, regressions and security before it ships.

• Review stage rejects diffs that delete existing behavior, ignore the task or contain obvious bugs
• Security review of AI-written output found and closed a path-traversal risk
• Generated tests run in a network-isolated sandbox
• Human approval required for any AI-chosen package; every step logged to an audit ledger