Herman Donawan
AI Developer & Programmer
Building low-latency AI agents & RAG systems that run 24/7 in production.
Profile
Herman Donawan
AI Developer & Programmer
- LocationBoyolali, Central Java, Indonesia
- Emailkangmaswawan21@gmail.com
- Local Time (WIB)--:--:--
About
AI Engineer building production LLM systems: RAG pipelines, RLHF evaluation frameworks, autonomous agents. Shipped voice-driven AI avatar serving 10k+ daily interactions. Full-stack (Next.js, FastAPI, Docker, Vercel/AWS).
3+
Years Experience
15+
Projects Built
3
Platforms
Focus areas
Featured Projects
Interactive AI Avatar / Radio DJ System
Real-time streaming app where a live AI avatar reads chat, responds to comments, and runs a 24/7 radio show. 10k+ daily interactions, <200ms latency.
Autonomous DeFi Liquidity Agent
Self-running agent analyzing DLMM liquidity positions on Solana, executing data-driven rebalancing. Hermes orchestration, on-chain execution.
Tech Stack
Engineering
AI & LLM
Voice & Audio
Experience
Freelance AI Contributor
2023 — PresentOutlier
- Designed RLHF evaluation pipeline for 50k+ coding samples; improved model pass@1 by 14% on HumanEval.
- Built automated quality gates reducing annotation cost 30% via prompt engineering & model-based eval.
- Curated multilingual audio datasets (EN/ID/JV) for voice AI training; 200+ hours validated.
Freelance Python & AI Developer
2023 — PresentUpwork
- Delivered 12+ client projects: LLM-powered automation (FastAPI + LangChain), voice AI pipelines (TTS + Whisper), DeFi analytics dashboards (React + Python).
- Maintained 5.0/5.0 rating across 15+ contracts; 100% on-time delivery.
- Designed autonomous agent architectures using Hermes for on-chain DeFi operations.
Community Ambassador
2023 — PresentBase Ecosystem
- Drove Web3 education & DeFi awareness across 5k+ community members; organized 8+ technical workshops.
- Fostered developer engagement: guided 50+ builders through Base onboarding, smart contract deployment.
Live SOL Price
RAG Playground
Ask me anything about documents...
Voice Agent
Tap to demo voice
More Projects
Data Science & EDA Portfolio
Collection of Kaggle notebooks: cleaning, exploration, feature engineering, model evaluation on real-world datasets. 15+ notebooks, 500+ views.
RAG SaaS — Multi-tenant Document Q&A
Voice Data Exploration: Gujarati Digits
Exploratory Data Analysis (EDA) on audio data, extracting key metrics like duration, sample rate, RMS energy, Signal-to-Noise Ratio (SNR), and Voice Activity Detection (VAD). Visualizes audio characteristics and quality.
By the Numbers
10k+
Daily Interactions
<200ms
Avg Latency
99.9%
Uptime
15+
Notebooks
Skills
Engineering
- Python
- TypeScript
- Next.js
- FastAPI
- Docker
- Git/GitHub
- CI/CD (GitHub Actions)
- PostgreSQL
- Redis
AI & LLM
- LLM Evaluation (RLHF)
- Prompt Engineering
- RAG Pipelines (LangChain, LlamaIndex)
- Vector Databases (Weaviate, Pinecone)
- Autonomous Agents (Hermes, CrewAI)
- Hermes Agent Framework
- Model Fine-tuning (LoRA/QLoRA)
- Model Serving (vLLM, TGI)
Voice & Audio
- Multilingual Voice Actor (EN / ID / JV)
- AI Voice Data Generation
- TTS Pipeline (Coqui, ElevenLabs, OpenAI)
- ASR (Whisper, Faster-Whisper)
- Real-time Voice Streaming
Available for new projects
Let's build something intelligent together.
I'm available for AI engineering, LLM evaluation, voice data, and automation projects. Questions, ideas, or just want to say hi — my inbox is open.