Approach 01 Selected Work 02 Experience 03 Download CV Portfolio Say Hello
AI Product Builder · Product Discovery · Rapid Prototyping

I turn ambiguous problems into products people can actually test.

Smallest useful MVPs · fast prototypes · real-user experiments. 1.6M+ installs across shipped games, plus published AI products and tools.

See selected work ↓ CV Portfolio LinkedIn GitHub
scroll

Most ideas die in decks. Mine become prototypes you can click.

How I work

A loop, not a checklist
01

Frame

User, constraints, business context — and the decision the product supports.

02

Scope

Core hypothesis first; the smallest honest test defines everything else.

03

Prototype

AI tools, APIs, lightweight builds — testable in days, not months.

04

Learn

Evidence changes direction faster than opinions do.

Selected work

Problems → experiments → decisions

Each project started with a real constraint, moved through evidence-driven trade-offs, and ended in a decision — not just a demo.

01 / 05
PP
AI Developer Tool · Published
1100 npm downloads since launch · 5 releases shipped in 4 days

Prompt Packer

problem → MVP scope → pricing → launch & feedback loops

  • Shipped an MCP server for Claude Code, Cline & Cursor — works with any MCP-compatible tool.
  • Full product ownership: problem definition → MVP scope → pricing → npm publishing → Telegram subscription flow.
  • Iterated on real usage: 5 versions in 4 days, MIT-licensed and public on GitHub.
MCP npm Telegram bot LLM APIs
Read case study ↗ npm ↗
02 / 05
AQ
AdTech · Experimentation
2× longer attention when users chose to engage · 93 real participants in a controlled A/B test

AdQuest

discovery → interaction design → live experiment → pivot

  • Tackled banner blindness with a voluntary-attention rewarded ad — no dark patterns, no coercion.
  • Ran the experiment, then killed the initial hypothesis after mapping the three-sided incentive problem.
  • Turned the pivot itself into a product decision: tiered reward model over a premature build cycle.
HTML5/JS Supabase Cloudflare Pages
Read case study ↗ Live demo ↗
03 / 05
RA
Computer Vision · Retail
CPU-only CPU-only · runs locally without GPU or cloud inference

Retail Analytics MVP

framing → prototype → failure analysis → iteration

  • Code runs explicitly on device='cpu' — no GPU server, no cloud, no inference bill.
  • Hand-wrote the track stitching: ghost IDs + distance matching + memory window — fixed double-counting behind shelves.
  • Privacy-first by design: video never leaves the store · EN/FA (RTL) interface, calibrated per shop.
YOLOv8n ByteTrack CPU Streamlit
Read case study ↗ GitHub ↗
04 / 05
GD
AI Safety · Product Infrastructure
sub-second risk-aware snapshot before a risky AI-agent action · Btrfs-native, zero downtime

Guardian

problem framing → risk model → prototype → technical validation

  • Built a risk-aware reverse proxy that intercepts AI-agent HTTP traffic and decides: snapshot or block.
  • Addressed the "undo" problem for autonomous agents — sub-second Btrfs snapshots as a safety net.
  • Designed for failure: if an agent breaks the system, recovery is one command.
Go Btrfs Reverse Proxy
View on GitHub ↗
05 / 05
SR
Product Infrastructure · Resilience
<500KB total client size · designed for constrained connectivity and 2G networks

Smart Defense Radar

architecture-as-product → build → constraint-driven iteration

  • Built for severe connectivity constraints: keeps the client lightweight and usable on slow networks.
  • <500KB client-side dashboard · Google Sheets as a zero-cost real-time database.
  • Proved infrastructure limits can become product requirements, not blockers.
Cloudflare Workers Activepieces Offline-First
Read case study ↗ Live ↗
keep scrolling →

Track record

Numbers behind the work
1.6M+ installs across 3 mobile games I helped build & ship
8 products built end-to-end — from problem framing to working prototype
93 real users in my controlled AdQuest A/B experiment
<500KB size of a crisis dashboard designed for constrained networks

Experience

From games to AI products
Nov 2025 — Present

AI Product Builder

Independent Product Work
  • Built 8 products end-to-end: AdQuest, Guardian, Prompt Packer, Retail Analytics, Smart Defense Radar.
  • Owned discovery → MVP scope → prototyping → validation and iteration based on evidence.
  • Built across AI developer tools, AdTech, computer vision, Product Infrastructure, and AI safety infrastructure.
Apr 2025 — Oct 2025

Founder · AI Product Builder

Arman Novaran Asr-e Digital · Alborz Science & Technology Park
  • Built and validated an AI customer-support SaaS concept, from problem discovery and user flows to technical validation.
  • Pivoted early when technical validation exposed security and multi-tenancy constraints — separating validation from a V1 build.
Mar 2021 — Feb 2024

Game Developer & Production Generalist

Buzzy Bee Game Studio
  • Programming, UI, animation and production across 3 shipped mobile games — 1.6M+ combined installs.
  • Progressed from C# programming intern to generalist production role owning cross-discipline delivery.

About

A non-linear path, one theme

Product first

I enjoy deciding what is worth building, why it matters, what to test first — and what evidence would change my mind.

Build to learn

I use AI tools, APIs and lightweight engineering to turn product hypotheses into working prototypes — fast enough to learn before committing to a full build.

Creative range

Games, AI tools, AdTech, retail analytics, crisis systems, automation and creative workflows — the common thread is turning ambiguous problems into testable products.

Currently

Deepening LLM product patterns, building new prototypes, and documenting the work in public.

Education

B.A. Illustration — Pars University of Art & Architecture
Associate Degree, Software Engineering
Vocational Diploma, Software Engineering

Languages & relocation

English — professional working proficiency · German — beginner, actively studying · French — elementary.
Open to relocation across Europe.

Open to AI Product Builder opportunities

Building something useful?

Looking to join an AI or early-stage product team where I can turn ambiguous ideas into working prototypes, validate them with real users, and iterate toward an MVP.

amirarmaniya@gmail.com LinkedIn GitHub Portfolio (PDF) Download CV

Based in Iran · Open to relocation across Europe