Senior Software Engineer

We are looking for a Senior Software Engineer to join our AI Centre of Excellence team at Epsilon, working at the heart of enterprise-grade Conversational, Agentic, and AI Platforms and applications. This role is for someone who brings a strong software engineering backbone, a data-oriented outlook, and deep hands-on expertise in Generative AI, RAG, and Agentic AI systems and is ready to help shape the future of intelligent, enterprise systems at scale.

You will be building and scaling AI-powered applications and agents that serve thousands of associates and clients across Epsilon and Publicis Groupe, integrating with enterprise systems. You will own and operate across the full lifecycle - from ideation, experimentation and prototyping to production hardening, evaluation, and operational governance.

The candidate will have responsibilities across the following functions:

Core Engineering and Architecture:

• Design, develop, and ship production-grade AI applications - including conversational Assistants, RAG pipelines, and multi-agent systems.
• Architect scalable, secure, and cost-efficient backend services using Python, Node.js, and cloud-native patterns (AWS / Azure/ GCP).
• Build and maintain API services (RESTful, streaming) that integrate AI capabilities with enterprise systems.
• Write clean, testable, well-documented code with CI/CD standards; champion engineering rigour in an AI-first team.

Generative AI and LLM Systems:

• Build and optimise LLM-powered features - including prompt engineering, structured output design, tool/function calling, and context management (multi-turn conversations, session handling).
• Design and implement evaluation frameworks (groundedness scoring, regression testing, quality benchmarking) for AI outputs - ensuring trust, accuracy, and continuous improvement.
• Stay hands-on with LLM APIs (Azure OpenAI, AWS Bedrock, Anthropic, open-source models) and make informed decisions on model selection, cost-latency tradeoffs, and fine-tuning vs. prompting strategies.

Retrieval-Augmented Generation (RAG):

• Design and build enterprise RAG pipelines, including embedding selection, chunking strategies, metadata enrichment, hybrid retrieval, re-ranking, and citation/traceability.
• Integrate and manage vector databases for scalable knowledge retrieval across heterogeneous enterprise data sources.
• Continuously improve retrieval quality by building golden test sets, measuring relevance, and implementing feedback loops.
• Work with Multimodal retrieval based on unstructured content.

Agentic AI and Orchestration:

• Design and implement agentic workflows - autonomous and semi-autonomous AI agents that can reason, plan, use tools, and implement multi-step business workflows with human-in-the-loop checkpoints.
• Build multi-agent orchestration frameworks using tools like AWS Bedrock, Agentcore, Cursor and other state-of-the-art open-source frameworks - enabling collaborative agent systems for complex enterprise scenarios.
• Develop reusable tool integrations that agents can invoke autonomously, with proper guardrails and safety controls.

Data and Analytics Mindset:

• Work with structured and unstructured enterprise data cleaning, transforming, and preparing data for AI consumption.
• Apply data science fundamentals (EDA, statistical analysis, anomaly detection) to diagnose issues, validate model behaviour, and derive actionable insights from AI system telemetry.
• Collaborate with data engineering teams to ensure data pipelines are reliable, timely, and aligned with AI feature needs.

Governance, Safety and Ops:

• Implement Responsible AI practices - including guardrails for hallucination handling, PII protection, restricted topic filtering, and compliance with enterprise security standards.
• Build and operate LLMOps / MLOps pipelines - model deployment, monitoring, logging, tracing, cost tracking, and lifecycle management.
• Contribute to SOPs, governance documentation, and operational runbooks for AI systems deployed across teams.

Requirements:

• Experience: 5-8+ years in software engineering, with at least 2+ years hands-on in Generative AI / LLM-based systems.
• Software Engineering: Strong proficiency in Python; experience with backend frameworks (FastAPI, Flask, Express/Node.js ); clean API design, version control (Git), testing, and CI/CD.
• Generative AI: Hands-on experience with LLM APIs (Azure OpenAI, AWS Bedrock, Anthropic, Google Gemini); prompt engineering, structured outputs, tool/function calling.
• RAG: Proven experience building RAG pipelines - embedding models, chunking, retrieval logic, vector database, re-ranking, and grounding.
• Agentic AI: Experience designing agent-based architectures - tool use, planning, multi-step workflows; familiarity with AWS Bedrock, Azure AI Foundry, or equivalent frameworks.
• Data Fundamentals: Solid understanding of data wrangling, SQL, EDA, and basic ML concepts; ability to work with structured/unstructured data at scale.
• Databricks: Designs, builds, and manages scalable data pipelines and AI solutions within the Databricks Lakehouse Platform.
• Cloud: Experience with AWS or Azure - deploying containerised services, serverless functions, and working with cloud AI/ML services.
• System Design: Ability to design distributed, scalable AI systems with clear tradeoffs on cost, latency, and reliability.

Good-to-Have:

• Multi-Agent Systems and A2A Protocols - experience with agent-to-agent communication patterns, Model Context Protocol (MCP), or similar emerging standards.
• Fine-Tuning and Model Adaptation - experience fine-tuning LLMs or adapter-based methods (LoRA, QLoRA) for domain-specific use cases.
• AI Evaluation and Benchmarking - experience building evaluation harnesses, automated grading, and regression testing for LLM outputs.
• Microsoft Ecosystem - familiarity with M365 Copilot, Copilot Studio, Bot Framework, Teams integrations, Adaptive Cards.
• Observability and Tracing - experience with AI-specific observability for debugging and monitoring AI systems in production.
• NLP and Classical ML - deeper grounding in NLP (named entity recognition, text classification, sentiment analysis) and classical ML (scikit-learn, XGBoost).
• Knowledge Graphs and Hybrid Search - experience combining graph-based retrieval with vector search for richer contextual grounding.
• Edge / Cost Optimisation - techniques for reducing inference cost, including model distillation, quantisation, caching, and batching strategies.
• Security and Compliance - awareness of data privacy regulations, secure API design, and AI red-teaming / adversarial testing.
• Bachelor's or Master's degree in Computer Science, Artificial Intelligence, Data Science, or a related field (or equivalent practical experience).