About this opportunity:
Join Ericsson as a Senior Software Architect – AI, owning end-to-end architecture for enterprise-scale GenAI and AI-powered solutions within our Self-Service Platform (SSP). You will design scalable, secure, production-grade AI platforms leveraging agentic frameworks, LLMs, and cloud-native services. You will establish reference architectures across RAG, memory, evaluation, and observability while guiding teams across the full model lifecycle — at the intersection of AI innovation and Responsible AI governance.
What you will do:
-
Architect agentic AI applications using LangChain, LangGraph, and orchestration patterns; define prompt strategies, guardrails, and structured outputs aligned to product and risk requirements.
-
Design and optimize RAG solutions (chunking, embeddings, retrieval, re-ranking) and own foundation model integrations (Azure OpenAI, AWS Bedrock, on-prem LLMs) with routing, fallbacks, and cost/performance optimization.
-
Define GenAI reference architectures; evaluate and select LLMs, embedding models, vector databases, and orchestration frameworks based on performance, compliance, and cost.
-
Embed security, privacy, and Responsible AI governance from inception — covering PII handling, data access controls, and content guardrails.
-
Build scalable backend APIs using Python (FastAPI, asyncio) with REST/JSON-RPC interfaces and resilience patterns (Redis, RabbitMQ); guide teams on MLOps/LLMOps standards including deployment, monitoring, retraining, and drift handling.
-
Define LLM evaluation strategies, implement observability/tracing (Arize, LangSmith), and design memory strategies with retention and replay safety for long-running assistants.
-
Containerize and deploy services via Docker and Kubernetes; govern CI/CD pipelines with automated testing, security scanning, and IaC (Terraform or equivalent).
The skills you bring:
-
BE/B.Tech/MCA in Computer Science, Engineering, or equivalent, with 15+ years in software architecture and relevant 3+ years designing AI/ML or LLM-based systems in production.
-
All academic credentials must be from recognized and accredited institutions and are further subject to verification.”
-
Deep expertise in Python (FastAPI, asyncio) and ML/DL frameworks (PyTorch, TensorFlow); strong experience with distributed, cloud-native services.
-
Hands-on with RAG pipelines, embeddings, and vector databases (Elastic, Pinecone, Milvus, Chroma) for enterprise knowledge grounding.
-
Hands on Python experience mandatory
-
Proven experience with agentic GenAI frameworks (LangChain, LangGraph, LlamaIndex, AutoGen) and interoperability patterns such as Model Context Protocol (MCP).
-
Strong knowledge of LLM architectures, fine-tuning techniques (LoRA, PEFT), and experience with Azure OpenAI and/or AWS Bedrock.
-
Solid understanding of MLOps/LLMOps, Responsible AI principles, and embedding governance into GenAI design.
Proficiency with Docker, Kubernetes, Terraform, and CI/CD for cloud-native AI deployments. -
Good to Have: LLM observability tools (Arize, LangSmith), Azure enterprise services (AKS, Key Vault), memory frameworks (MemGPT, LangMem), knowledge graph experience, and Telecom industry AI adoption background.
-
Locations: Bangalore, Kolkata, Gurgaon, Noida, Chennai
