AI Engineer - Autonomous Agents & Model Infrastructure

About The Role

We're seeking an experienced AI Engineer to design, deploy, and manage autonomous agent systems on proprietary infrastructure. You'll own the full lifecycle—from optimizing model weights to building production-grade agents with fine-tuning and reinforcement learning on on-premises or private cloud environments.

Key Responsibilities

• Design and deploy autonomous agent architectures on AWS VPC and on-premise environments
• Manage model weights and optimize for inference; implement LoRA and QLoRA fine-tuning for domain-specific tasks
• Develop reinforcement learning pipelines for agent training with reward modeling and policy optimization
• Implement MLOps/LLMOps infrastructure: model versioning, A/B testing, rollbacks, and evaluation frameworks
• Architect RAG systems integrating vector databases with proprietary and fine-tuned models
• Optimize model serving infrastructure (vLLM, TorchServe, TensorRT) for production inference
• Build monitoring and observability systems for agent behavior and RL training quality
• Ensure model security, data privacy, and audit compliance in enterprise deployments

• 3-4 years hands-on experience in AI/ML/Data Science with at least 2 projects shipped to production
• 2+ years dedicated experience in MLOps, LLMOps, or AIops (model deployment, inference optimization, pipeline automation, model management)
• AWS proficiency across AI services: EC2, VPC, S3, IAM, SageMaker, Bedrock, Lambda, or custom ML infrastructure
• Strong software engineering fundamentals: containerization (Docker), orchestration (Kubernetes), CI/CD, and API design
• Hands-on experience deploying and serving large language models or foundation models in production environments
• Practical experience with LoRA (Low-Rank Adaptation) and QLoRA (Quantized LoRA) fine-tuning techniques for efficient model adaptation
• Understanding of reinforcement learning fundamentals and experience implementing RL-based training: policy gradients, reward shaping, or preference-based optimization
• Working knowledge of vector databases and RAG implementation
• Solid understanding of model optimization techniques and inference constraints (GPU memory, latency, throughput)

• Experience building autonomous agents with RL frameworks (DPO, PPO, RLHF) and fine-tuning frameworks (Hugging Face Transformers, PEFT)
• QLoRA experience on consumer-grade GPUs in memory-constrained environments
• Migration experience from cloud APIs (OpenAI, Anthropic) to self-hosted models
• On-premises or VPC-only deployment experience
• Familiarity with agent frameworks (LangChain, LlamaIndex, AutoGen) and MLOps tools (MLflow, W&B, DVC)
• Strong debugging and systems thinking approach with evidence-based problem-solving

AWS (EC2, VPC, S3, IAM, SageMaker), Python, Docker, Kubernetes, LLMs/Foundation Models, LoRA/QLoRA Libraries (PEFT), Reinforcement Learning Frameworks, Vector Databases, Agent Frameworks, Model Serving Technologies, Fine-tuning Tools (Hugging Face Transformers), CI/CD Pipelines, PyTorch/TensorFlow

What We Value

Pragmatic engineers who balance performance and cost, strong debuggers with evidence-based approaches, clear communicators, and owners who see projects end-to-end.

Application Requirements

Submit GitHub portfolio with production ML systems, MLOps implementations, fine-tuning work (LoRA/QLoRA), and RL-based agent training examples. Bonus: on-premises deployments or model optimization projects.