Hashmat R

AI/ML intern

JFF Consultants • Tiempo completo

May 2026 - Jul 2026 • 2 mos

Developed and contributed to production-grade AI document-processing systems for quality assurance workflows, processing scanned documents and PDFs through OCR, extraction, validation, and structured output stages. – Built end-to-end text extraction and processing pipelines using PaddleOCR, PP-Structure, EasyOCR, Gemini OCR, OpenCV, and PyMuPDF to extract text, layout, and structured information from complex documents. – Designed LangGraph-based workflows to orchestrate document ingestion, OCR, LLM-based extraction, normalization, validation, retry handling, and final output generation. – Implemented a complete Human-in-the-Loop (HITL) workflow that routes low-confidence or failed extractions for human validation before producing final structured results. – Used Pydantic schemas to enforce structured LLM outputs, validate extracted fields, and maintain reliable data contracts throughout the processing pipeline. – Developed RAG components using embeddings and vector databases to retrieve relevant context from custom knowledge sources and improve context-aware LLM responses. – Worked with LangChain and LLM APIs to implement intelligent information extraction, retrieval, reasoning, and document-processing components. – Containerized AI services using Docker and worked with production-oriented application environments, focusing on modular architecture, error handling, retries, validation, and system reliability.