AI Engineer focused on multimodal models, agentic workflows, and Document AI. I build retrieval-augmented and vision-language systems that combine RAG, VLMs, and tool-driven automation to handle complex enterprise docume…
AI Engineer focused on multimodal models, agentic workflows, and Document AI. I build retrieval-augmented and vision-language systems that combine RAG, VLMs, and tool-driven automation to handle complex enterprise document workflows. Experienced in designing scalable ML pipelines—from data processing to model optimization and serving—supported by solid MLOps practices. Committed to delivering reliable, high-performance document intelligence solutions that reduce manual work and create measurable operational impact.
Designed and deployed multimodal and vision-language models (VLMs) to enhance document understanding and improve robustness across production workflows.
Integrated retrieval-augmented generation (RAG) with VLMs and LLMs to ground responses in enterprise data and reduce hallucination in user-facing applications.
Built evaluation frameworks for multimodal and retrieval pipelines, including accuracy benchmarking, hallucination analysis, and continuous quality monitoring.
Developed hybrid retrieval pipelines—dense, sparse, and cross-encoder reranking—to improve search relevance and responsiveness for large-scale document repositories.
Implemented advanced Document AI components such as layout analysis, table extraction, signature detection, and cross-modal validation for high-accuracy IDP pipelines.
Engineered scalable data pipelines for text, OCR outputs, images, and multilingual datasets, ensuring consistent normalization and efficient ingestion into training and retrieval systems.
Owned the full ML lifecycle and MLOps stack, from experimentation to optimization and deployment, using Docker, Kubernetes, CI/CD, and monitoring tools to maintain production-ready models.
Designed and deployed multimodal and vision-language models (VLMs) to enhance document understanding and improve robustness across production workflows. Integrated retrieval-augmented generation (RAG) with VLMs and LLMs to ground responses in enterprise data and reduce hallucination in user-facing applications. Built evaluation frameworks for multimodal and retrieval pipelines, including accuracy benchmarking, hallucination analysis, and continuous quality monitoring. Developed hybrid retrieval pipelines—dense, sparse, and cross-encoder reranking—to improve search relevance and responsiveness for large-scale document repositories. Implemented advanced Document AI components such as layout analysis, table extraction, signature detection, and cross-modal validation for high-accuracy IDP pipelines. Engineered scalable data pipelines for text, OCR outputs, images, and multilingual datasets, ensuring consistent normalization and efficient ingestion into training and retrieval systems. Owned the full ML lifecycle and MLOps stack, from experimentation to optimization and deployment, using Docker, Kubernetes, CI/CD, and monitoring tools to maintain production-ready models.
Designed ML and DL pipelines for unstructured document processing, improving the reliability and consistency of extraction and classification workflows.
Enhanced OCR performance by developing preprocessing and augmentation strategies tailored to noisy and low-quality document images.
Built computer vision models for structured data extraction and document understanding, with a focus on robustness across diverse layouts and image conditions.
Integrated multimodal visual question-answering components to automate insight extraction and reduce reliance on manual review.
Contributed to an end-to-end document automation system combining classification, extraction, and validation modules to streamline operational workflows.
Designed ML and DL pipelines for unstructured document processing, improving the reliability and consistency of extraction and classification workflows. Enhanced OCR performance by developing preprocessing and augmentation strategies tailored to noisy and low-quality document images. Built computer vision models for structured data extraction and document understanding, with a focus on robustness across diverse layouts and image conditions. Integrated multimodal visual question-answering components to automate insight extraction and reduce reliance on manual review. Contributed to an end-to-end document automation system combining classification, extraction, and validation modules to streamline operational workflows.
The AI-first Recruitment Operating System that unifies and automates the entire hiring lifecycle for modern enterprises worldwide.
Dr Job FZ LLC. 2026 © All Rights Reserved
