Web Development
- Full build, design to deploy
- Admin panel — edit content yourself
- SEO + analytics included
2–6 weeksStart a project
Fullstack & AI builderJakarta · ID
Portfolio · 2026
Building the systems your business runs on.
I’m Valentino Allen, an independent fullstack engineer and AI builder. I design, build, and ship the software teams rely on day-to-day — web platforms, AI automations, and WhatsApp booking systems — from first scope call to live deploy.
Jakarta · GMT+7 · remote worldwideNext.js · TypeScript · n8n · Claude2 storefronts live · 2 AI tools built
One engineer, end to end: I take projects from scope call to live deploy — no agency layers, no handoffs.
I work with international teams and Indonesian businesses that need software actually shipped: storefronts that replace manual catalog sharing, automations that remove repetitive work, booking flows that run themselves on WhatsApp. Alongside client work I’m a computer science student, and I build my own ML tools — which is why the demos further down this page run for real.
Local time--:--:--Jakarta, Indonesia · remote worldwide
Focus
Web platforms, AI automation, booking systems, and WhatsApp AI — built and maintained by the same person you scoped them with.
- 2024Started a Computer Science degree; first builds shipped alongside coursework.
- 2026Two client websites live in production; two ML tools built — the demos further down this page run for real.
- NowFreelance fullstack and AI work with international teams, while finishing the degree.
Here's how I can help.
Four ways teams hire me. Every engagement is scoped and priced up front, and each one ends with something running in production — not a prototype.
AI Automation
- Workflow mapped from your process
- n8n build + your integrations
- Monitoring + handoff docs
1–3 weeksStart a project
Booking Systems
- Flow matched to your rules
- Admin dashboard for schedules
- Auto confirmations + reminders
2–4 weeksStart a project
WhatsApp AI Booking
- Natural-language booking in WhatsApp
- Calendar + database integration
- Human handoff built in
2–4 weeksStart a project
What working together looks like.
The same four steps on every project, whether it’s a two-week automation or a full storefront. You always know what’s happening and what you get next.
Scope
A short call plus a written brief. We pin down what actually needs to exist, what doesn't, and what done looks like.
You getA fixed-scope proposal with timeline and price.
Build
I build on a staging URL you can open from day one. Progress updates as links to the real thing, not slide decks.
You getA working preview you can click, every week.
Ship
Deploy on your domain with SEO, analytics, and performance checks done. I walk your team through running it.
You getProduction software, live, with a handoff walkthrough.
Support
Launch isn't the end. I stay available for fixes and small adjustments while the work settles into daily use.
You get30 days of post-launch fixes included.
I reply to every inquiry within 24 hours on weekdays.
Self-built tools, running live.
Two ML tools I built to fix problems the businesses I work with actually have. The demos below aren’t videos — they run as you scroll.
raw review1 of 1,500
Kualitasnya oke, tapi barang sampai dalam keadaan peyok.
ABSA
NLP · Aspect-Based Sentiment AnalysisPulls a separate opinion per aspect out of messy, code-switched Indonesian reviews — not one blunt star rating.
It scores five aspects — product quality, price, shipping, seller service, packaging — across four classes including not-mentioned. The hard part it's built for: implicit aspects, where the sentiment is there but no keyword names the topic (“arrived dented” is packaging, negative).
Best of 4 models benchmarked — IndoBERT, macro-F1 ≈ 0.86, ≈ 0.90 recall on implicit aspects.
PythonTransformersStreamlit
Built on 1,500 labelled Tokopedia reviewsFootfit
Computer VisionEstimates shoe size from one photo, using a bank card in frame as a ruler.
Wrong-size returns are a constant cost for online shoe sellers. Footfit measures a foot against a known-size reference card, converts pixels to centimetres, and maps that to a size — before the order, not after the return.
Computer VisionPython
Open live demo
Estimate the right size before the order ships.
- card85.6 mm known
- scale1 px = 0.42 mm
- foot26.4 cm
Shipped, not just demoed.
Both of these are live businesses you can visit right now — built solo, from design to deploy, for owners who needed their catalog on the web, not in a chat thread.
Nata Teknik
Hardware retail brought online. A live product catalog and SEO-ready storefront for a Bali-based hardware store, replacing manual WhatsApp catalog sharing with a searchable web presence.
Live in production — the team browses, searches, and shares the catalog from one URL instead of resending photos on WhatsApp.
Visit natateknik.com
Pradipa Indonesia
Leather goods, brought online. Storefront for an Indonesian leather brand selling handcrafted cardholders and MagSafe accessories — clean product pages, fast on mobile, ready for direct-to-consumer sales.
Live in production — product pages the brand updates itself, fast enough on mobile to sell direct-to-consumer.
Visit pradipaindonesia.comWhat clients say about working with me.
From the owners of the two storefronts above — in their words, in both languages we worked in.
0102
Have a project? Let’s build it.
Tell me what you’re building and where it should run. The more concrete the brief, the faster the reply.
Or reach out directly