Jinu Augustine

Ai Solutions Architect @Demetrix Infotech

Toronto, ON, CA

Architected and deployed domain-specific RAG architectures, leveraging large language models (LLMs) such as Mistral Instruct, Llama 3.1, Qwen 2.5 and vector databases such as Pinecone, Qdrant and FAISS- Led a team of engineers in packaging AI-based solutions as web applications throughout the software development lifecycle- Fine-tuned and deployed LLMs on Azure and AWS, including multimodal models, to deliver tailored solutions for various applications- Developed advanced search and retrieval systems using techniques such as hybrid search and contextual compression techniques and frameworks such as Langchain and Llama Index to enhance information retrieval accuracy- Deployed LLMs at scale utilizing inference engines like vLLM and quantization methods to optimize performance and achieving latency while mitigating hallucinations- Successfully implemented AI-driven compliance tools that reduced regulatory response times by 70% for government agencies and NGOs- Implemented robust security measures to ensure compliance with company-specific guidelines- Utilized tools like Llama Guard to build guardrails and mitigate potential risks- Established performance metrics for faithfulness, context relevance, and ground-truth based evaluation- Collaborated with subject matter experts to refine model outputs and ensure alignment with user expectations.

Acharya Institute of Technology

Bachelor of Engineering

Durham College

Post Graduate Certificate