Machine Learning Engineer

We're looking for an experienced ML Operations Engineer to design, scale, and continuously improve our AWS-based ML platform powering a Contract Intelligence solution.

This role goes beyond traditional MLOps—you will actively contribute to model experimentation, evaluation, and optimization for extracting structured insights from large, complex legal documents.

You will work to improve data extraction accuracy, clause identification, and risk detection, while optimizing processing performance at scale.

Key Responsibilities

ML Engineering & MLOps

• Design, deploy, and scale ML pipelines using tools like MLflow, SageMaker, Kubeflow, Airflow, or DVC
• Implement CI/CD for ML models including versioning, automated testing, deployment, and rollback
• Build and maintain robust training and inference pipelines on AWS

Document AI & NLP Systems

• Work on large, dense document processing (contracts, legal documents, scanned files)
• Enhance existing OCR and parsing pipelines (already built) to improve accuracy and efficiency
• Develop and optimize models for:
• Data extraction
• Clause extraction
• Risk identification

Model Development & Optimization

• Fine-tune open-source LLMs and NLP models (LLaMA, Mistral, Falcon, BERT variants)
• Continuously update prompts and fine-tune models for better performance
• Optimize models for latency, cost, and scalability (quantization, batching, etc.)

Experimentation & Research

• Run structured experiments including:
• A/B testing
• Error analysis
• Model benchmarking
• Research and evaluate new open-source models and techniques
• Drive improvements in accuracy and processing time through iterative experimentation

Data & Training Strategy

• Work with existing training datasets to:
• Improve data quality
• Build and refine training pipelines
• Support client-specific model customization and fine-tuning
• Create feedback loops for continuous learning and model improvement

Production & Monitoring

• Build scalable pipelines using Lambda, ECS/Fargate, and Step Functions
• Implement monitoring using CloudWatch and custom dashboards
• Track model performance using domain-specific metrics
• Manage model lifecycle using SageMaker Model Registry

Required Qualifications

• 3+ years in ML Engineering / MLOps with production NLP systems
• Strong Python skills with frameworks like PyTorch, Hugging Face, spaCy
• Hands-on experience with:
• SageMaker (training, endpoints, pipelines)
• CI/CD for ML systems
• Containerization (Docker)
• Experience with:
• Fine-tuning and deploying open-source LLMs
• Large document processing pipelines (PDF, Word, scanned docs)
• Model evaluation and optimization
• Strong understanding of:
• Experimentation (A/B testing, error analysis)
• Model performance tuning and trade-offs

Preferred Qualifications

• Experience in industries such as:
• LegalTech
• Insurance
• Pharma
• BFSI
• Familiarity with:
• Document AI pipelines (OCR, parsing, entity extraction)
• RAG architectures and vector databases
• Parameter-efficient fine-tuning (LoRA, QLoRA)
• Exposure to:
• Multi-GPU / distributed training
• Domain-specific NLP models (e.g., Legal-BERT)

Tech Stack

ML & NLP: PyTorch, Hugging Face, spaCy, LangChain

Models: LLaMA, Mistral, Falcon, BERT variants

AWS: SageMaker, Lambda, ECS/Fargate, Step Functions, Textract

Data & Storage: S3, DynamoDB, Aurora PostgreSQL

Vector Search: OpenSearch, pgvector

MLOps: MLflow, SageMaker Pipelines, Model Registry

Monitoring: CloudWatch, Prometheus, Grafana

IaC: Terraform, CloudFormation