AI Engineer: Document Intelligence & Machine Learning

EarlySafe works with complex financial and lending information including **financial statements, bank statements, GST information, property documents, legal reports, valuation reports, verification reports, scanned PDFs, tables and images**. We are looking for an AI/ML Engineer who has **built and deployed ML or Document AI systems in production** and can transform unstructured information into reliable, structured evidence. **What You Will Do** - Build production **Document AI** pipelines - Develop OCR, layout understanding, table extraction and document classification - Use **VLMs and multimodal models** for document understanding - Extract entities, values and evidence from unstructured information - Develop ML models for classification, anomaly detection, similarity and pattern recognition - Reconcile information across documents and data sources - Build confidence scoring, validation and exception handling - Develop training, inference and evaluation pipelines - Monitor model accuracy, drift and production failures - Build scalable APIs and services consumed by EarlySafe applications **Production Experience Required** You should have taken ML/AI models beyond notebooks and POCs into **real production environments**. You should understand: **Training/Inference Pipelines • Model Serving • APIs • Docker • Cloud Deployment • CI/CD • Model Versioning • Monitoring • Data/Model Drift • Batch Processing • Scalability • Failure Handling** Experience designing for **large documents, high-volume processing and asynchronous workloads** is particularly valuable. **Skills We Value** **Core:** Python • Machine Learning • Document AI • VLMs • NLP/Information Extraction • OCR/Layout Understanding • Production ML **Useful:** PyTorch/TensorFlow • Hugging Face • Scikit-learn • OpenCV • XGBoost/LightGBM • FastAPI • MLflow • Spark/Databricks • Docker/Kubernetes • AWS/Azure/GCP **What We Are Looking For** We want engineers who care about **accuracy, confidence, provenance, reconciliation and reproducibility**. The question isnt simply whether the model returned an answer. It is whether that answer can be **trusted at enterprise scale**. Experience with financial documents, banking, lending, fintech or Intelligent Document Processing is particularly valuable.