Machine Learning Engineer (GenAI / LLM)

Intelligent Image Management Inc (IIMI) is an IT Services company reimagines and digitizes data through document automation using modern, cloud-native app development. As one of the world's leading multinational IT services companies with offices in the USA and Singapore, India, Sri Lanka, Bangladesh, Nepal and Kenya. Over 4,200 people are employed by IIMI worldwide whose mission is to advance data process automation. US and European Fortune 500 companies are among our clients.

Become part of a team that puts its people first. Founded in 1996, Intelligent Image Management Inc. has always believed in its people. We strive to foster an environment where all feel welcome, supported, and empowered to be innovative and reach their full potential.

Website: https://www.iimdirect.com/

About the Role:

We are seeking an experienced Machine Learning Engineer (GenAI / LLM) to lead the development of intelligent document processing (IDP) systems specializing in historical and archival records. In this role, you will design end-to-end computer vision and LLM/VLM pipelines to restore degraded scans, transcribe complex handwriting (HTR), and extract structured, schema-validated data from records such as census ledgers, birth, marriage, and death certificates. You will leverage modern OCR/HTR models, fine-tune vision-language architectures, and deploy scalable cloud-based inference pipelines to turn complex, unstructured historical archives into reliable, searchable databases.

Key Responsibilities

• Archival ML Pipelines: Architect, train, and deploy robust, multi-stage pipelines tailored for historical document processing—spanning ingestion, preprocessing, binarization, de-skewing, line/word segmentation, transcription, and structured entity extraction.
• Handwritten Text Recognition (HTR) & Layout Analysis: Design and fine-tune computer vision and deep learning models for layout analysis, table extraction, and handwriting transcription across diverse scripts, historical cursive styles, varied ink qualities, and tabular forms.
• Document Image Restoration: Apply advanced computer vision techniques (using OpenCV, PyTorch, etc.) to enhance low-contrast, bleed-through, degraded, or faded historical scans while strictly preserving resolution, archival integrity, and metadata.
• OCR/HTR Engine Integration: Benchmark, integrate, and customize managed document AI tools (AWS Textract, Google Cloud Document AI / Vision API) alongside open-source OCR/HTR frameworks (e.g., TrOCR, PaddleOCR, Kraken, Tesseract) to maximize character/word-level accuracy on non-standard typography and handwriting.
• LLM & Multimodal (VLM) Extraction: Implement open- and closed-source LLMs/VLMs (e.g., Gemini, Claude, GPT-4o, LLaMA, Mistral, Qwen-VL) to transcribe, normalize, correct OCR/HTR errors, and extract schema-adherent fields (names, dates, relationships, locations, causes of death, occupations) from unstructured or semi-structured records.
• Entity Linking & Structured Output Generation: Develop reliable post-processing pipelines using Named Entity Recognition (NER), regex validation, and JSON/Pydantic schemas to output standardized, audit-ready data suitable for database ingestion and genealogical/client analytics.
• MLOps, Serving & Scaling: Build, monitor, and scale distributed, multi-threaded batch and streaming inference pipelines using tools like MLflow, Docker, Ray, AWS SageMaker, or GCP Vertex AI to efficiently process massive collections of high-resolution image archives.
• Quality Assurance: Define CER/WER metrics, entity-level validation, schema validation pass rates) and human-in-the-loop (HITL) active learning workflows to track accuracy and refine model confidence.

Required Qualifications

• Education & Experience: Bachelor’s or Master’s in Computer Science, Data Science, or related quantitative field, with 5+ years of production ML/CV experience.
• Computer Vision & HTR: Hands-on experience with layout analysis, object detection, and deep-learning-based handwriting recognition (TrOCR, ViT, CRNN).
• Applied LLM/VLM Skills: Direct experience fine-tuning, prompting, and serving LLMs/VLMs for structured data extraction and reasoning over complex documents.
• Core Technical Stack: Advanced Python proficiency using PyTorch, OpenCV, Hugging Face Transformers, and distributed data processing libraries.
• Cloud & MLOps: Practical experience with AWS or GCP ML stacks (SageMaker, Vertex AI), containerization (Docker), model serving, and lifecycle tracking.

Preferred Skills

• Practical experience with historical census, vital records, or dense tabular document parsing.
• Hands-on knowledge of Multimodal RAG, vector databases, and spatial bounding-box indexing.
• Experience generating synthetic handwriting datasets or mentoring junior ML engineers.

Work Location: Work from Office

Send cover letter, complete resume, and references to email: tech.jobs@iimdirect.com

Industry:

• Outsourcing/Offshoring

Employment Type

Full-time