# Tanveer Riaz - Curious mind. Builder mode. Tanveer is an AI specialist in Singapore focused on agentic systems. Curiosity is how he works: he follows questions until they become things people can open, inspect, and learn from. His perspective combines two decades of global banking systems delivery with public experiments in AI agents, local models, design, and useful Singapore products. He is active in Singapore’s AI builder community and publishes lessons from public sessions and industry events with the builds they changed. Location: Singapore Last updated: 2026-08-02 ## Public work ### [SideSpark](https://github.com/tanveerriaz/SideSpark) Hackathon prototype · Public prototype A playful way for coworkers to meet beyond their usual circles through a short Skill Swap or Social Connect break. Evidence: Both choices lead from picking a break to meeting a clearly labelled demo sidekick, completing a shared activity, and saving a small takeaway on the device. Tags: Workplace connection, Skill sharing, Shared breaks, ChatGPT Sites ### [Chroa](https://chroa.vercel.app) Building now · Public beta A colour-system studio that generates accessible OKLCH palettes, wires them into live interface previews, and exports a DESIGN.md a coding agent can follow. Evidence: Five live preview types, measured contrast, and agent-ready design output. Tags: Design systems, Accessibility, Coding agents, React ### [Skillz](https://github.com/tanveerriaz/Skillz) Open source · Public A public library of reusable AI-agent skills: focused procedures that turn prompt experiments into repeatable execution. Evidence: 26 published skills, cross-agent conventions, security-first guidance, MIT licensed. Tags: Agent skills, Open source, AI workflows, Python ### [HalalEats SG](https://www.halaleats-sg.com/) Singapore product · Live A chat-first halal makan companion for Singapore, designed around plain-language questions and MUIS-certified venue data. Evidence: Public, anonymous product with local search, venue discovery, and practical Singapore context. Tags: RAG, Singapore, Halal tech, FastAPI, Postgres ### [Nexus Bank](https://banking-demo-production.up.railway.app/) Fictional banking demo · Live demo A fictional AI-banking interface used to explore how conversational assistance, financial context, and bounded agent actions might fit together. Evidence: A live public interface using mock data only; it is a learning demo, not a financial service. Tags: Fictional demo, Banking UX, Agentic systems, Mock data ### [PromptZap](https://promptzap.io/) Browser tool · Live A browser extension that turns rough instructions into structured prompts inside the AI tools people already use. Evidence: One-click workflow with 23 prompt frameworks across ChatGPT, Claude, Gemini, and more. Tags: Browser extension, Prompt design, AI tooling ### [ScriptToReel](https://github.com/tanveerriaz/scripttoreel) Open-source pipeline · Public A local-first Python pipeline that researches, scripts, assembles, narrates, and renders short videos from a topic. Evidence: CLI and local dashboard over the same pipeline, with explicit checkpoints and reproducible project folders. Tags: Python, Video, Local workflow, FFmpeg ### [Markdown Convert](https://markdown-ext.pages.dev/) Agent utility · Live A one-click browser extension that converts webpages into clean Markdown for agents, notes, and knowledge bases. Evidence: No account, no backend, no build step; handles readable content, tables, forms, and useful embedded regions. Tags: Markdown, Browser extension, Agent-ready content, JavaScript ### [Codex Pet League](/blog/dashboard-atlas) Data experiment · Build story A playful leaderboard for tiny coding companions - and a build story about replacing persuasive mock data with verifiable public signals. Evidence: The useful part is the correction: provenance labels, real public data, and an explicit scoring model. Tags: Data provenance, Dashboard, Game, Build in public ### [Logo Challenge](https://github.com/tanveerriaz/Logo-Challenge-Game) Voice game · Public prototype A voice-powered brand guessing game built as a fast family-friendly experiment in multimodal interaction. Evidence: A compact public prototype that combines speech input, progressive levels, hints, and feedback. Tags: Voice UI, Game, Rapid prototyping ## Choose a starting point ### I am new to AI Start with the concepts, then open one small working example. Path: Plain-language glossary → PromptZap → Field notes ### I want to build with agents See how prompts become skills, harnesses, and verifiable workflows. Path: PromptZap → Skillz → Chroa ### I work in enterprise Connect agent experiments to risk, provenance, and human ownership. Path: Banking perspective → Nexus Bank → AI governance credentials ### I want local and private AI Explore what can run on-device and where cloud models still help. Path: Local-first AI → Ollama + MLX → ScriptToReel ## Curiosity trails ### Can better instructions become better visual stories? Still unfolding · PromptZap → Chroa → ScriptToReel A rough-prompt utility opened a deeper question about visual taste, agent-readable design systems, and eventually reproducible media pipelines. ### Is the newest model the answer - or is the harness? Deepening · multi-model experiments → coding agents → reusable skills Comparing models led to a stronger conclusion: tools, permissions, context, feedback loops, and repeatable skills often matter more than a leaderboard position. ### How much useful AI can stay on my own machine? Prototyping · OpenRouter → Ollama → MLX / oMLX Cloud routing makes comparison easy; local inference tests privacy, control, latency, and resilience against real workflows instead of slogans. ### What changes when agent ideas meet banking constraints? Applying experience · global systems delivery → Nexus Bank demo → human review Traceable data, bounded actions, explicit hand-offs, and clear stopping points matter when agent ideas meet banking constraints. ### Can a very local problem become a useful product? Live · Singapore context → verified venue data → HalalEats SG A focused local need became a chat-first discovery product grounded in verified public data and the way people actually ask where to makan. ### What can a game teach that a dashboard cannot? Playground · voice game → pet leaderboard → provenance lesson Play makes feedback immediate. It also exposes when delightful presentation outruns the evidence underneath it. ## Learning now ### Agent systems Deepening: How tools, memory, permissions, and evaluation turn a model into a dependable workflow. ### Local inference Prototyping: What can stay on-device without making the product slower or less useful. ### AI-readable publishing Applying here: How HTML, Markdown, JSON-LD, and explicit content policies can make a personal site easier for agents to understand. ## Content policy Public personal work only. Views are the author's own. No secrets, credentials, local file paths, exact location, private repositories, private agent configuration, employer-confidential material, or client data is included. ## Structured endpoints - [Complete site JSON](/data/site.json) - [Projects JSON](/projects.json) - [Starting points JSON](/start-here.json) - [Curiosity trails JSON](/journeys.json) - [Glossary JSON](/glossary.json) - [Events JSON](/events.json) - [Tools JSON](/tools.json) - [Shipping signals JSON](/signals.json) - [Blog JSON](/blog/index.json) - [LLM guide](/llms.txt)