Skip to main content
Esa Krissaesa krissa

Building useful software from Bali

Hi, I'm Esa Krissa.

I build backend systems and AI that connects to the real world.

Practical APIs, agent workflows, and automation for travel, hospitality, and the teams behind them.

Explore my work Get in touch
Bali, IndonesiaGitHub LinkedIn

A little of what I build

Useful AI starts with real work and dependable tools.

01Guest request
Agent workflow
03Connected tools
Hospitality · AI · OperationsBali, Indonesia

Engineering with a purpose

Good software solves the problem in front of someone.

I'm Esa, a backend and AI engineer based in Bali. My work sits where dependable services meet language models, connected tools, and real business workflows.

I care about the parts that make a system useful after the demo: clear API boundaries, human approval when it matters, sensible integrations, and software that teams can actually rely on.

Recently, that means building for hospitality and travel—domains I know from home—while exploring agentic systems, cloud services, and the infrastructure behind them.

More about my work on GitHub

A few things I’ve built

Real projects at the intersection of AI, APIs, and everyday operations.

An AI-powered operations assistant for villa and hotel managers in Bali, combining agent workflows with the tools a real hospitality team needs.

PythonFastAPILangGraphModel Context Protocol
In progress
1Live site

An MCP server that connects hotel search workflows to Booking.com, making live accommodation search available to AI agents as a tool.

PythonModel Context Protocol
On GitHub
299

A Cognito-authenticated user service built around a DynamoDB single-table design, Lambda, and EventBridge for service-to-service events.

TypeScriptAWSServerless FrameworkDynamoDB
On GitHub
Live site

A LangGraph booking flow that pauses for human approval at the right moment instead of treating every decision as fully automated.

PythonLangGraph
On GitHub
1

A WhatsApp chatbot powered by GPT-5 Mini, LangChain, and Twilio—bringing an LLM workflow into a familiar messaging channel.

PythonLangChain
On GitHub

A small multi-agent system with separate agents for coding, data, and research tasks.

PythonLangGraph
On GitHub
1

An example LangChain agent using MCP adapters to connect its reasoning to external tools.

PythonLangChainModel Context Protocol
On GitHub
4
Explore all my repositories

Tools for the work

A flexible toolkit for building dependable services and connected AI workflows.

Languages

Python
Core
TypeScript
Regular
JavaScript
Regular
Go
Exploring
Rust
Exploring

Backend & APIs

FastAPI
Core
Node.js
Regular
Next.js
Regular
Serverless Framework
Regular

AI & LLM Agents

LangChain
Core
LangGraph
Core
Model Context Protocol
Core
OpenAI API
Core
Anthropic API
Core

Cloud & DevOps

AWS
Core
Google Cloud Platform
Core
Docker
Core
Terraform
Regular
GitHub Actions
Core

Databases

PostgreSQL
Core
Redis
Regular
Supabase
Regular
MongoDB
Regular
DynamoDB
Regular

Dev Tools

Cursor IDE
Core
Claude Code
Core

Python for the engine room

Python is my primary language for APIs, automation, and agent workflows. I reach for TypeScript when a product or service calls for it, from AWS serverless systems to thoughtful web interfaces.

See more tools & technologies

Databases

PostgreSQLRedisMongoDBSupabaseDynamoDB

Cloud & delivery

AWSGoogle CloudDockerTerraformGitHub Actions

Languages & tools

TypeScriptJavaScriptGoRustCursorClaude Code

Research & writing

Academic work that connects ideas about AI systems with working implementations.

JournalJuly 2025
First authorBuletin Teknologi Informasi

AI-Based Travel Agent Architecture with Implementation of LangGraph and Model Context Protocol (MCP)

Multi-agent AI system using Large Language Models (LLM) for an integrated travel agent application managing hotel bookings, flights, and tour packages with responsive customer service. Implemented LangGraph framework for multi-agent coordination using StateGraph and cyclical graphs. Applied Model Context Protocol (MCP) for standardized AI assistant integration with external data sources. Performance: 99.7% uptime, response time 6.6-53.6 seconds, error rate <0.5%, accuracy rate 92-96%.

LangGraphMCPMulti-agent AITravel Booking
ConferenceSeptember 2025
Engineering Proceedings (MDPI)

N-Gram and Full-Text Search Algorithm Testing for Pattern Recognition in a Chatbot Engine

Research on the ISONER (Information System On Internet Messenger) framework, focusing on pattern recognition services using Natural Language Processing (NLP). Implemented and benchmarked N-gram and Full-Text Search algorithms using MySQL's built-in plugins for low-latency pattern matching in chatbot engines. The FTS and four-gram algorithms showed excellent results.

N-GramFull-Text SearchChatbotNLP
Read the paper

Have something in mind?

I'm always happy to talk through a thoughtful project or a good engineering problem.

Looking for someone to build dependable backend systems, connect an AI workflow to your tools, or simply trade ideas? Drop me a note.

Say hello
Emailesakrissa.wayan@gmail.comGitHubgithub.com/esakrissaLinkedInlinkedin.com/in/esakrissa