Lead AI & Deep Learning Engineer

EPAM Systems; United States; On-site; USD 170,000 - 210,000; Full time

Job summary

EPAM Systems Inc. seeks a Lead AI & Deep Learning Engineer to bridge traditional ML/DL with agentic AI patterns for real-time workflows.

You will deliver production-ready agentic systems for content scanning, tagging, and contextual ad insertion, while designing event-driven microservices over WebSockets with OpenSearch and Weights & Biases evaluation. Responsibilities include designing agentic AI frameworks, building ML/DL models for OCR and audio analysis, and maintaining observability

Qualifications

• Experience with traditional ML/DL and agentic AI workflows.
• Hands-on with event-driven microservices and MCP over WebSockets.

Responsibilities

• Design and deploy agentic AI frameworks including context management and event-driven architectures.
• Fine-tune ML/DL models for classification, OCR, audio analysis.
• Architect search and retrieval systems using vector databases like OpenSearch/Elasticsearch.
• Implement evaluation pipelines with Weights & Biases to validate models before production.
• Collaborate in a fast-paced Kanban setup with rapid prototyping and reviews.
• Deliver autonomous workflows over WebSockets for live streaming tagging.

Skills

Python programming

FastAPI

Agentic AI workflows

MCP

WebSockets

Tools

OpenSearch/Elasticsearch

Weights & Biases

LangChain

LlamaIndex

Terraform

AWS Cloud

Job description

We are seeking a Lead AI & Deep Learning Engineer to connect classic ML/deep learning with agentic AI patterns for real-time live-stream workflows. You will deliver production-ready agentic systems for content scanning, tagging, audio/visual classification, and contextual ad insertion, while designing event-driven microservices over WebSockets with vector search/OpenSearch and Weights & Biases evaluation.

Responsibilities

• Design and deploy advanced agentic AI frameworks, including context management, MCP over WebSockets, and event-driven architectures for autonomous enterprise workflows
• Build and fine-tune traditional ML and deep learning models for classification, OCR (Optical Character Recognition), and audio spectrogram analysis
• Architect and optimize high-performance search and retrieval systems using OpenSearch / Elasticsearch and vector databases
• Implement rigorous evaluation pipelines in Weights & Biases (or equivalent ML/LLMOps tools) to validate new model versions against evals before production switching
• Collaborate in a fast-paced, experimental Kanban setup with daily discussions, technical reviews, and rapid prototyping
• Deliver robust autonomous agentic workflows over WebSockets that handle content scanning and live streaming tagging without hallucinations
• Integrate high-accuracy classification, OCR, and audio processing models into production MURAL and MURAL LIVE pipelines
• Maintain comprehensive evaluation and observability dashboards in Weights & Biases governing model updates and agentic tool calls

Requirements

• Proven track record with 5+ years of software engineering experience spanning traditional machine learning/deep learning (classification, OCR, computer vision/audio) and modern agentic AI workflows
• Demonstrated experience building stateful, event-driven microservices and agentic frameworks (MCP, WebSockets, tool-call orchestration)
• Hands-on experience with search and vector database infrastructure (OpenSearch / Elasticsearch)
• Strong experience with model evaluation, tracking, and experimentation tools (Weights & Biases)
• High proficiency in Python & FastAPI backend development
• Deep understanding of agentic AI frameworks such as LangChain, LlamaIndex, or custom agents
• English proficiency at B2 level or higher
• Nice to have: Experience with fine-tuning, distillation, and quantization of open-source models
• Domain background in media & entertainment (broadcast video supply chain, Peacock / live streaming architecture)
• Familiarity with Go / gRPC microservices
• Knowledge of Terraform & AWS Cloud infrastructure