AI Engineer& Full StackDeveloper
IbuildLLMsystemsthatworkreliablyinproduction.MyfocusisgroundinglanguagemodelsinlivebusinessdatathroughRAG,fine-tuningwithQLoRAfordomainaccuracy,andshippingagenticpipelineswithanti-hallucinationguarantees.
$ ananta --profile
name·Ananta Viryadiva
role·AI Engineer
status·● available
spec·Anti-Hallucination Arch.
method·QLoRA · PEFT · 4-bit
models·LLaMA 3 · Claude · Ollama
stack·Python · RAG · n8n
projects·4 production systems
clients·Textilindo · Surya Toto
Ecsconstruksi
$
Featured Projects
Production-grade solutions leveraging state-of-the-art AI stacks.
const onte = createAssistant({
channels: ['WhatsApp', 'Telegram'],
memory: 'Redis',
calendar: 'Google Calendar',
});
await onte.connect();
// ✓ WhatsApp connected
// ✓ Telegram ready
// ✓ Calendar synced
Project 01
In DevelopmentO.N.T.E – Personal AI Ecosystem
Personal Project
A fully automated personal AI assistant connecting WhatsApp, Telegram, and Google Calendar with human-in-the-loop approval, anti-detection heuristics, and Redis-backed context memory.
Challenge
Orchestrating multi-platform AI flows with human approval without sacrificing response naturalness
Impact
Zero-intervention scheduling and messaging with human-in-the-loop control, running daily as a personal tool
Project 02
AI Customer Service System
PT Textilindo
End-to-end AI customer service system combining a fine-tuned LLM, a dual-knowledge-base RAG pipeline, and a live ERP integration layer — designed so the model never fabricates pricing, stock, or product data.
Challenge
Grounding a generative model in volatile, real-time business data without hallucination
Impact
A fully automated CS pipeline where every price, stock figure, and product variant is sourced live — the model cannot fabricate what it does not know
Project 03
Calibration Management System
PT Surya Toto Indonesia
Web-based calibration tracking system for production equipment with RBAC, automated reminders, and an immutable audit trail for compliance.
Challenge
Manual calibration tracking with no audit trail or compliance readiness
Impact
Reduced manual input, improved traceability and compliance readiness for QA & Maintenance
const app = createWebApp({
stack: ['Tailwind', 'MySQL'],
features: ['RBAC', 'Reporting'],
client: 'PT Ecsconstruksi',
});
await app.deploy();
// ✓ Production-ready
// ✓ Stakeholder-approved
// ✓ Delivered on time
Project 04
Internal Web App Suite
PT Ecsconstruksi
End-to-end internal web applications for a construction company, delivered as sole developer with rapid stakeholder-aligned delivery.
Challenge
Rapid delivery with cross-functional stakeholder alignment across departments
Impact
Responsive, production-ready systems delivered on time with zero post-launch rework
Technical Arsenal
LLM Models & Platforms
LLaMA 3
Meta
Gemini
Qwen
Alibaba AI
Claude
Anthropic
Ollama
Local LLM
OpenClaw
Web Intel
AI / LLM
Full Stack
Engineering
Tools & Infra
Get in Touch
Available for freelance and contract work in AI engineering and full-stack development. Reach out through any of the channels below.
$ agent --status
My AI Agent is currently
under development.
When live, this chat connects to my personal AI Agent running on my own infrastructure. You can ask anything about my projects, technical decisions, or work experience and get real answers from the same agent I use for day-to-day automation.
Personal AI Agent
Built on a live agentic system running on my own infrastructure. Not a scripted FAQ — it actually knows my work.
Deep project knowledge
Ask specific technical questions about architecture decisions, stack choices, and implementation details from any of my projects.
Same agent, multiple platforms
The same agent that powers my WhatsApp and Telegram automation, accessible here as a single interface on the portfolio.
Context-aware conversation
Supports multi-turn conversations with rolling context memory, so follow-up questions work naturally without losing track.
Active when AI Agent is connected