This is a submission for the New Year, New You Portfolio Challenge Presented by Google AI
About Me
I am a full-stack software engineer based in South Africa, focused on building durable software that ships and lasts.
I care about clear system boundaries, low-maintenance architectures, and products that solve real problems without unnecessary complexity. Most of my work lives in production: SaaS tools, automation systems, payment flows, and community platforms.
This portfolio is meant to reflect how I actually work. Pragmatic, intentional, and biased toward execution.
Portfolio
You can also visit it at: https://smngvlkz.com
This portfolio is structured like internal system documentation rather than a traditional marketing site. It highlights:
- Live and archived products I have shipped
- Clear product intent, status, and scope
- Ongoing research and exploratory work
- SaaS, automation, payments, and platform work
- Open-source and community contributions
AI Integration
Design principle:
AI is used only where it improves inspectability or reduces cognitive load. All core functionality remains deterministic and debuggable without AI.
This portfolio includes an interactive terminal (SYSTEM.QUERY) powered by Google Gemini 3 Flash.
Users can query the entire portfolio data directly:
Structured commands (handled locally for speed):
- help - list available commands
- list all - show all products and contributions
- show activity - show GitHub & GitLab stats, streak, sessions
- show [id] [field] - show specific data for any item in the portfolio
- explain [id] - show full breakdown for any product or contribution
- list fields - show queryable fields
Natural language queries (powered by Gemini) - examples:
- "what tech does paychasers use?"
- "compare the infrastructure of all products."
- "which product uses blockchain?"
- "what's the Cape Community Blog built with?"
The system prompt constrains Gemini to portfolio data only, with terminal-style output formatting. Common queries are handled locally without an API call. Complex or natural language queries fall back to Gemini.
This demonstrates practical AI integration: fast where possible, intelligent when needed, without over-engineering.
Live Activity Data (GitHub & GitLab)
SYSTEM.ACTIVITY pulls live data from GitHub and GitLab using authenticated API requests.
A server-side route aggregates commits, repositories, contribution streaks, and session metadata into a unified activity model. This model is used in two places:
Rendered as live system activity stats in the interface, including a 111-day contribution heatmap that visualizes real contributions with interactive tooltips (showing source breakdown: GitHub/GitLab commits, date, and day of week). Only days with actual contributions are inspectable on hover.
Exposed to SYSTEM.QUERY, allowing users to inspect and query the same data via the terminal.
This keeps activity up to date without manual updates and demonstrates real external API integration alongside AI-powered querying, with a single source of truth shared across UI and terminal
How I Built It
- Framework: Next.js
- Language: Typescript, JavaScript
- Styling: Tailwind CSS
- AI: Google Gemini 3 Flash
- Deployment: Google Cloud Run (fully managed, container-based)
I used AI-assisted development to iterate on structure, copy clarity, and information hierarchy, while keeping the final implementation intentionally simple and deterministic.
The site favors static rendering, fast load times, and a small surface area for long-term maintainability. The terminal feature adds interactivity without compromising the minimal aesthetic.
What I'm Most Proud Of
- The restraint in the design. Nothing exists without a reason
- Clear labeling of product status (live, inactive, archived)
- Showing real shipped work instead of demo projects
- AI integration that fits the product's identity (terminal queries, not chatbot fluff)
- A portfolio that reflects engineering maturity, not trend-chasing
This is not a highlight reel. It is a snapshot of how I build software today.
Top comments (0)