Ai Engineer

## About the role We are looking for a highly technical AI Engineer to architect and build the foundational AI infrastructure for our next-generation Smart Audit platform. In this role, you will hold end-to-end ownership of the AI lifecycle—moving beyond third-party API orchestration to design, train, and deploy proprietary machine learning models and customized Large Language Model (LLM) architectures natively. You will transform how the enterprise analyzes financial records, contracts, and operational data, turning manual, error-prone sampling into automated, 100% population risk scanning. If you are passionate about custom architecture, enterprise scalability, and deploying high-performance systems to production, this role is for you. ## Key Responsibilities - Proprietary Model Architecture and Training: Design, train, and fine-tune domain-specific LLMs and deep learning models for anomaly detection, fraud identification, and automated compliance checking. - Custom GenAI and Retrieval Infrastructure: Engineer advanced, low-latency retrieval systems, custom embedding models, and multi-agent workflows optimized for massive volumes of unstructured legal and financial data, independent of high-level orchestration frameworks. - Enterprise Production and MLOps: Establish automated, end-to-end MLOps pipelines covering data labeling, continuous training, version control, and high-throughput/low-latency model serving in production environments. - Data Architecture Engineering: Architect custom ETL and data ingestion pipelines capable of parsing and structuring complex, multi-format enterprise data (unstructured PDFs, massive Excel sheets, ERP system dumps). - Evaluation and Robustness Frameworks: Implement deterministic evaluation frameworks to strictly benchmark model accuracy, mitigate hallucinations, and minimize false positives/negatives in a high-stakes auditing environment. - Security, Privacy and Explainability: Architect systems to comply with strict enterprise data isolation and privacy standards (e.g., GDPR, SOC2). Ensure models provide clear, explainable lineage so human auditors can verify the AI's logical reasoning. ## Technical Skills and Qualifications ### Required - Experience: Demonstrated track record of building, scaling, and maintaining machine learning models in a live enterprise production environment. - Deep Learning and Frameworks: Professional proficiency with Python and deep learning frameworks (such as PyTorch or TensorFlow) along with a deep mathematical understanding of neural network architectures. - Custom LLM Development: Proven experience with parameter-efficient fine-tuning (PEFT, LoRA/QLoRA), deep speed optimization, and custom alignment (RLHF/DPO) of open-source models (e.g., Llama, Mistral). - Infrastructure and MLOps: Practical experience with containerization (Docker, Kubernetes) and MLOps tooling (Triton Inference Server, vLLM, MLflow, Ray) for optimizing inference speeds and infrastructure costs. - Advanced Document AI: Expertise in customized NLP, Named Entity Recognition (NER), and building custom OCR/Document parsing pipelines for highly unstructured documents. - Vector Infrastructure: Experience designing and optimizing distributed Vector Databases (e.g., Qdrant, Milvus, pgvector) at scale. ### Preferred (Nice-to-Have) - Experience building and scaling AI solutions within Enterprise SaaS or highly regulated industries (Finance, Fintech, Legaltech). - Experience with Knowledge Graphs or Graph Neural Networks (GNNs) for mapping complex corporate entities and transactions. - Contributions to open-source AI/ML libraries or published research in NLP/Anomaly Detection. ## Soft Skills and Culture Fit - First-Principles Thinker: You look past black-box solutions; you understand how a model behaves at the architectural level and how to optimize it via data quality and hyperparameter tuning. - System Ownership: Comfortable owning the entire technical roadmap from initial design to highly available production systems. - Pragmatic Innovator: Ability to balance cutting-edge research with the practical realities of enterprise engineering constraints (cost, latency, and security).