01 / The Problem
AI product engineering tidak berhenti di prompt.
Bagian tersulit dari AI app ada di area sekitar LLM: data flow, UI state, auth, validation, streaming, tool execution boundary, logging, deployment, dan cost control.
AI Product Engineering with TypeScript Enrollment Closed — Next Batch Coming
10 Weeks AI Product Engineering with TypeScript
Program live-class untuk membangun product system di sekitar LLM, bukan hanya memanggil model API. Student akan belajar menghubungkan product surface, backend, data layer, AI harness, streaming UI, tool calling, observability, dan deployment dalam satu workflow TypeScript.
Dimulai
13 June 2026
Kelas live setiap Sabtu dan Minggu pukul 10.00 WIB.
Program Flyer
Download the AI Product Engineering flyer.
Download PDF ringkasan program ini untuk melihat arah pembelajaran, jalur product engineering, perjalanan kurikulum, profil mentor, harga, dan detail pendaftaran.
Download PDF02 / The Strategy
This program takes the product engineering path.
AI Engineering punya dua jalur besar. Model engineering fokus pada dataset, training, dan eval model. Program ini fokus pada product engineering: membuat LLM menjadi fitur produk yang bisa dipakai, dijaga, dan dikembangkan.
03 / The System
Calling an LLM API hanya satu layer dari produk.
Student tidak langsung lompat ke inference. Mereka membangun fondasi produk lebih dulu, lalu menambahkan AI layer dengan kontrak, state, permission, observability, dan feedback loop.
The product system around the LLM.
01
Product Surface
Reactive UI, streaming states, generative interfaces, and user-facing flows.
02
Reliable Backend
Hono APIs, Zod contracts, durable objects, permissions, and workflow boundaries.
03
Data Layers
PostgreSQL, vector embeddings, user-owned data, and application context.
04
AI Harness
AI SDK, tools, observability, evals, cost control, and deployment infrastructure.
A 10-week journey from web foundation to reliable AI product.
Week 1
Typed Web Foundation
Modern JavaScript review, runtime mental model, and package management
TypeScript fundamentals, interfaces, generics, and type design
Project setup with Vite, PNPM, linting, formatting, and Git workflow for product teams
Week 2
Reliable Backend with Hono
HTTP fundamentals, routing, handlers, middleware, and error boundaries
Request validation with Zod and typed API contracts
Building backend boundaries that are ready for frontend, data, and AI features
Week 3
Data Layers for AI Products
PostgreSQL fundamentals, schema design, relations, and migrations
Prisma workflow for queries, transactions, and maintainable data access
Designing data models for user-owned content, AI outputs, and product workflows
Week 4
Product Surface with React
React components, forms, routing, and state boundaries
TanStack Query for server state, caching, mutation flow, and loading states
Building product screens that stay responsive while backend and AI work happens
Week 5
Product Readiness
Authentication, authorization, session handling, and protected routes
File uploads, object storage, and user-owned resources
Environment variables, deployment workflow, permissions, and basic observability
Week 6
AI Harness and SDK Integration
OpenAI JavaScript/TypeScript SDK and provider-agnostic AI SDK patterns
Prompt design, structured output, schema validation, and fallback handling
Designing LLM calls as product features, not isolated prompt demos
Week 7
Streaming UI and Generative Interfaces
Streaming text responses from API routes into React UI
Chat UX, message state, optimistic UI, and failure recovery
Generative UI patterns for summaries, drafts, copilots, and review flows
Week 8
Tools, Agents, and Workflow Boundaries
Tool design, tool input schemas, approval flow, and safe execution boundaries
Connecting AI features to database operations and internal services
Building agentic workflows with logs, retries, and clear user feedback
Week 9
Retrieval, Evaluation, and Reliability
Document ingestion, embeddings, retrieval flow, and practical RAG architecture
Evaluation harness, test prompts, regressions, and quality checks
Cost control, rate limits, model selection, and abuse prevention
Week 10
AI Product Engineering Final Project
Build a production-minded AI product with a typed fullstack foundation
Present architecture, tradeoffs, and deployment decisions
Final Assignment Briefs
The order of the curriculum may change, but everything listed in the curriculum will be learned.
What students should be able to do after the program.
Typed Fullstack Skill
Build frontend, backend, API contracts, and database access in one coherent TypeScript workflow.
AI Product Engineering
Build product surfaces around LLMs with streaming UI, structured output, tool calling, and workflow boundaries.
Production Judgment
Understand auth, permissions, deployment, observability, cost control, and failure states for AI features.
Indra Zulfi Mushoddaq
Interdisciplinary Software & AI Engineer
- AI Engineering Lead @ AlphaX, Japan
- Prev. Designstripe Canada, Generalist and Software Engineer
- Prev. SSH International Bahrain, Design Team
Affordable pricing with comprehensive features and lifetime access to course materials
Enrollment Closed — Next Batch Coming
20 Live Class Sessions
2x 1-on-1 Mentoring
Auto Assignment Reviews
Lifetime Recording Access
Bonus Platform Course
Completion Certificate
Everything you need to know about the program structure.
Schedule
Saturday and Sunday - 10.00 WIB
Starts Saturday, 13 June 2026
Duration
10 Weeks
2 live sessions per week
Prerequisites
Basic HTML, CSS, and JavaScript
TypeScript will be taught from the foundation
Format
100% Online Live Interactive Classes
Live class, assignment, review, and mentoring
Final Output
AI product engineering project
A deployable fullstack product for portfolio and review
Stack
React, Hono, PostgreSQL, TypeScript
With AI SDK, Zod, Prisma, and deployment workflow
Why this direction fits Devscale.
One Language Across the Product
TypeScript lets students share types, schemas, and product assumptions across frontend, backend, data, and AI harness code.
Built for AI Product Engineering
The curriculum focuses on the hard parts around the model: UI state, streaming, tools, data access, auth, and deployment.
Modern Web Stack
Students work with React, Hono, PostgreSQL, Prisma, Zod, TanStack Query, and AI SDK patterns used in current product teams.
Mentor-Led Feedback
Live classes, assignments, and 1-on-1 sessions give students direct feedback on architecture, code quality, and product decisions.
Support system to keep students moving.
Discord Community
Learn alongside Devscale students and alumni in a focused engineering community.
Assignment Feedback
Submit work, receive feedback, and improve through practical iteration instead of passive watching.
Mentor Guidance
Use 1-on-1 sessions to discuss technical blockers, project direction, and career context.
Related Reading
The thinking behind this program direction.
Career direction
Two Paths for AI Engineers
Kenapa program ini fokus ke LLM product harness, bukan model training dari nol.
Read the article
Program direction
Menutup Program AI-Enabled Python Web Development
Alasan Devscale mengalihkan fokus program publik dari Python ke full-stack TypeScript.
Read the article
Common questions before joining the program.
Apakah program ini menggantikan AI-Enabled Python? ⌃
Untuk program publik Devscale, iya. AI-Enabled Python kami sunset agar fokus kurikulum berikutnya lebih tajam ke AI Product Engineering dengan TypeScript. Python tetap bagus dan tetap relevan, terutama untuk data, automation, machine learning, dan eksperimen model. Namun untuk membangun produk web modern dengan UI, backend, database, auth, streaming, dan AI feature dalam satu workflow, kami memilih TypeScript sebagai fokus utama.
Apakah saya harus sudah menguasai TypeScript? ⌃
Tidak harus. Kamu perlu nyaman dengan dasar HTML, CSS, dan JavaScript terlebih dahulu. TypeScript akan dipelajari dari fondasi, mulai dari type dasar, interface, narrowing, generics, sampai cara memakai type untuk menjaga API contract, schema, dan data flow di aplikasi fullstack.
Apakah program ini cocok untuk pemula? ⌃
Cocok untuk beginner yang sudah punya dasar web development. Program ini bukan mulai dari nol absolut, jadi kamu sebaiknya sudah memahami struktur HTML, styling CSS, JavaScript dasar, function, array, object, dan async dasar. Jika belum, kamu tetap bisa mendaftar, tapi kami sarankan menyelesaikan materi dasar web terlebih dahulu sebelum batch dimulai.
Apakah pembayaran bisa 2x? ⌃
Program ini biasanya tersedia dengan opsi pembayaran 2x. Informasi lebih lanjut akan diumumkan saat pendaftaran batch berikutnya dibuka.
Apakah akan belajar AI model training? ⌃
Tidak. Fokus program ini bukan melatih model dari nol. Fokusnya adalah membangun produk yang mengintegrasikan AI secara utuh: memakai LLM API, AI SDK, structured output, streaming UI, tool calling, workflow automation, retrieval, evaluation sederhana, dan production safety. Jadi arahnya adalah AI Product Engineering, bukan machine learning research.
Apa final project yang akan dibuat? ⌃
Final project berupa aplikasi fullstack TypeScript yang punya fitur AI nyata. Student akan membangun product surface untuk single-turn AI interaction, multi-turn agent experience, agentic workflow dengan tool calling, RAG atau retrieval layer, serta observability untuk melihat trace, logs, error, dan kualitas output. Yang dinilai bukan hanya AI-nya, tapi juga arsitektur aplikasi, database design, UX, auth, deployment, error handling, dan keputusan teknisnya.