Overview
Senior Associate AI Engineer in PwC's Advisory line of service (Data, Analytics & AI specialism), designing and operating ML pipelines, LLM serving and LLMOps for client solutions.
What you'll do
- Design ML pipelines for experiment management, model management, feature management, and model retraining.
- Design APIs for model inferencing at scale.
- Achieve better latencies and accuracies in model results.
- Reduce training and resource requirements for fine-tuning LLM and LVM models.
- Apply distributed training and serving of large language models, including model and data parallel training with DeepSpeed and service frameworks such as vLLM.
- Apply model fine-tuning and optimization techniques.
- Apply DevOps and LLMOps practices, including Kubernetes, Docker, and container orchestration.
- Work with LLM orchestration frameworks such as Flowise, Langflow, and Langgraph.
What you'll need
- Bachelor of Technology (B.Tech), MCA, BCA or M.Tech.
- 5-8 years of experience.
- Proven expertise with MLflow, SageMaker, Vertex AI, and Azure AI.
- Deep knowledge of GPU architectures.
- Expertise in distributed training and serving of large language models.
- Proven expertise in model fine-tuning and optimization.
- Proven expertise in DevOps and LLMOps practices.
- Knowledge of Kubernetes, Docker, and container orchestration.
- Deep understanding of LLM orchestration frameworks such as Flowise, Langflow, and Langgraph.
- Mandatory skills: Gen AI, LLM, Huggingface, Python, PyTorch/TensorFlow/Keras, Langchain, Langgraph, Docker, Kubernetes.
- Required skill: Project Management.
- Proficient in Python, SQL and JavaScript.
Nice to have
- AWS Professional Solution Architect certification (bonus).
- AWS Machine Learning Specialty certification (bonus).
- Azure Solutions Architect Expert certification (bonus).
Details
- Location: Chennai - Menon Eternity.
- Visa sponsorship: No.
- Government clearance required: No.
- Job posting end date: September 20, 2026.
Read the full description and apply on the company’s own careers page.