Software Engineer - building production AI systems.
I've spent 6+ years building fullstack web and mobile applications. These days I'm focused on the intersection of LLMs, vector search, and retrieval-augmented generation - shipping AI features that actually work in production, not just demos.
- 🔭 Currently building: AI Agent Starter Kit - production RAG pipelines, tool-calling agents, and MCP server patterns
- 🌐 Portfolio: mamenesia.com
- ✍️ Writing: blog.mamenesia.com
- 📫 Contact: dedy.prasetyo.h@gmail.com
| Layer | Tools |
|---|---|
| Languages | TypeScript, JavaScript, Rust, Go, Python |
| Frontend | React, Next.js, Svelte, Astro, Tailwind CSS |
| Backend | Node.js, Express, NestJS, REST & GraphQL APIs |
| Mobile | React Native, Flutter |
| AI / ML | LangChain, LangGraph, OpenAI, Pinecone, RAG, prompt engineering |
| Data | PostgreSQL, MySQL, Firebase, Prisma ORM |
| Infra | Docker, GitHub Actions, CI/CD pipelines |
AI Agent Starter Kit Production-ready starter kit for AI agents. RAG pipeline with ingestion, semantic search, cited answers, and hallucination detection. Built-in evaluation harness, LangFuse observability, cost tracking, and exponential backoff retry. Not a toy.
Pokedex GraphQL Fullstack Pokémon encyclopedia built with Next.js 15, GraphQL (Apollo Client), TypeScript, and Tailwind CSS. Advanced search, multi-type filtering, side-by-side Pokémon comparison, and comprehensive test coverage.
- AI Analytics Platform - Built semantic search using OpenAI embeddings + Pinecone, enabling contextual discovery across large document sets
- Enterprise LMS - Architected fullstack Next.js learning platform at Kalbe Farma; delivered MVP in 2 months (ahead of 3-month schedule)
- Insurance Automation - Reduced claims processing from 3 days to 1 day through workflow automation (66% faster)
- 10+ projects delivered across logistics, insurance, e-commerce, and edtech - maintaining 100% client satisfaction
