AI Engineer · LLM Systems · Agentic AI
Building intelligent systems that scale.
Production-grade LLM engineering, RAG pipelines, and multi-agent orchestration — delivered remotely to teams worldwide. From enterprise communication platforms to low-resource Sinhala NLP, I build AI that ships and performs.
01 / About
I'm Amalsha Fernando — a remote AI/ML Engineer specializing in large language model systems, retrieval-augmented generation, and multi-agent orchestration. My work sits at the intersection of rigorous engineering and emergent AI capabilities.
At EchonLabs, I architect agentic pipelines for Sinhala language processing and have built enterprise-scale communication platforms handling 10,000+ daily interactions across voice, email, and messaging. I care deeply about AI observability, hallucination mitigation, and pushing the boundaries of what production LLM systems can reliably do.
With a BSc in Information Technology from SLIIT, a peer-reviewed publication in autonomous systems, and a self-compiled Sinhala TTS dataset — a rare asset in South Asian NLP — I bring both academic depth and practical delivery to every engagement.
Remote-first by design. My UTC+5:30 timezone works naturally with US afternoon, EU morning, and APAC schedules — no timezone friction.
02 / Skills
03 / Experience
04 / Projects
Enterprise agentic AI system orchestrating 10,000+ daily interactions across voice, email, and messaging. LangGraph-based orchestration with persistent memory — 95% STT accuracy and 40% faster response times in production.
HIPAA/GDPR-compliant clinical NLP summarization using LoRA fine-tuned BART. OCR extraction, structured section parsing, and PII anonymization — deployed for sensitive healthcare data.
Full-stack AI platform with FastAPI backend, async task execution, and geo-specific proxy rotation. Advanced user behaviour simulation at scale — built for real-world production load.
05 / Education & Credentials
Peer-reviewed publication in autonomous drone systems. Active research in LLM architectures and agentic AI. Self-compiled Sinhala TTS dataset — a unique asset for South Asian AI development, sought after by regional NLP teams.
Certifications
Open to AI engineering roles, research collaborations, and consulting contracts. Remote-first. Available to overlap with US, EU, and APAC business hours.