Shakyadissayake

Associate Machine Learning Engineer

Augustory Corp • Full time

Dec 2025 - Present • 5 mos

1. Collaborated with the Qimana (USA) engineering team to architect and deploy scalable AI systems in a distributed product environment, contributing to end-to-end feature delivery across research, backend, and data engineering workflows. 2. Engineered an Agentic AI framework integrating Knowledge Graphs and Graph RAG using Neo4j, enabling multi-hop reasoning over structured and unstructured enterprise data and improving contextual response relevance by ~30–40% compared to baseline vector-only retrieval. 3. Co-developed a real-time voice-to-voice intelligent interviewer agent using OpenAI APIs and WebSocket-based streaming, reducing response latency by ~35% and enabling seamless bidirectional conversational interactions for dynamic candidate assessment. 4. Designed and implemented a production-ready document extraction and ingestion pipeline leveraging Gemini Embeddings, LangChain semantic chunking, and agentic chunking with narrative-based story generation; optimized graph ingestion into Neo4j, increasing retrieval accuracy and reducing redundant node creation by ~25%. 5. Contributed to the architectural design and schema modeling of a FinTech Knowledge Graph for ledger account management, enabling structured financial relationship modeling, faster query execution (~40% improvement), and improved traceability of transaction flows across accounts. 6. Improved overall system robustness through iterative prompt engineering, retrieval optimization, and evaluation benchmarking, ensuring higher factual grounding and reduced hallucination rates in LLM-generated outputs.