Senior AI/ML Architect

Role: Senior AI/ML Architect

Location: SFO, CA (Onsite)

Employment Type: Full-Time with Criticalriver

Experience: 10–15 Years (4+ years ML/DL, 2+ years LLM/GenAI)

Job Summary

We are seeking a highly skilled Senior AI / ML Architect with strong hands-on experience in LLM training, fine-tuning, cloud deployment, model hosting, and MLOps. The ideal candidate will lead the end-to-end lifecycle of AI models, from data preparation and model selection to deployment, monitoring, and optimization in production environments.

Key Responsibilities

• Design and implement AI/ML solutions for training and fine-tuning LLMs using enterprise data.

• Build end-to-end pipelines covering data preparation, model training, evaluation, deployment, and monitoring.

• Perform hands-on LoRA/QLoRA fine-tuning and optimize training parameters.

• Develop scalable GPU-based training environments on AWS, Azure, or GCP.

• Deploy and host models using vLLM, TensorRT-LLM, SGLang, or TGI.

• Establish MLOps pipelines using MLflow, W&B, Docker, Kubernetes, Ray, and CI/CD tools.

• Monitor model performance, infrastructure utilization, latency, throughput, and costs.

• Troubleshoot training, inference, deployment, and production issues.

• Provide technical leadership and mentor AI/ML engineering teams. Mandatory Skills

1. LLM Training & Fine-Tuning

• PyTorch, Hugging Face Transformers

• LoRA, QLoRA, PEFT, TRL, Accelerate

• Experience fine-tuning LLMs such as Llama, Qwen, Mistral/Mixtral, Gemma, DeepSeek, or Phi

• Strong understanding of datasets, hyperparameters, and GPU optimization

2. Model Hosting & Cloud Deployment

• AWS, Azure, or GCP

• GPU infrastructure (A100, H100, L40S)

• Docker, Kubernetes

• vLLM, SGLang, TensorRT-LLM, TGI

• Production-grade model serving architectures

3. MLOps & Lifecycle Management

• MLflow, W&B

• Model registry, versioning, monitoring, rollback

• CI/CD for ML pipelines

• Automated training, deployment, and retraining workflows

Preferred Skills

• Python, SQL

• Data engineering and large-scale data pipelines

• Spark, Databricks

• Quantization techniques (GPTQ, AWQ, bitsandbytes, FP8)

• RAG, Vector Databases, Agentic AI, MCP, Tool Calling

• LLM evaluation and regression testing frameworks