Lumen Technologies
Website:
lumen.com
Job details:
Key Responsibilities
- Solutioning using new technologies with a fail-fast, learn-fast mindset; own PoCs and pilot deployments.
- Architect composable pipelines across LLMs and core ML; design reusable components (RAG, embeddings, prompt chains).
- Lead technical discovery, map high-impact use cases, and translate them into modular designs.
- Implement robust MLOps: standardized experiment tracking, model registries, automated CI/CD, and Kubernetes-based serving.
- Integrate vector stores, retrieval services, and policy/guardrails for safety and compliance.
- Establish observability baselines (tracing, metrics, drift/alignment dashboards) for both agents and models.
- Perform advanced agent evaluation (task success, tool-use correctness, groundedness, hallucination checks) and RAG evals.
- Collaborate with product and platform teams to drive adoption and handover to production.
- Experience in quality assurance solutioning and automation tools exposure would be an added advantage
Skills And Tools Required
- Advanced Python; strong with PyTorch/TensorFlow, scikit-learn, Hugging Face ecosystems.
- Experience with LangChain, Semantic Kernel, LangGraph, AutoGen.
- Working knowledge of MCP and A2A communication patterns; ability to design basic multi-agent workflows.
- Cloud AI services: Azure AI, Vertex AI, AWS SageMaker; containerization with Docker/K8s.
- MLOps toolchain: MLflow/Weights & Biases, DVC, GitHub Actions/Azure DevOps, Feast (feature store).
- Observability stack: LangSmith, PromptFlow, OpenTelemetry, Evidently AI, Great Expectations.
Preferred Qualifications
- Bachelor’s or Master’s in CS/AI/DS.
- Experience with vector databases (FAISS, Pinecone, Milvus, Redis), and evaluation frameworks (OpenAI Evals, custom harnesses).
- Exposure to domain solutions (e.g., Telecom OSS/BSS, NOC analytics, customer service automation) is a plus.
Natural Language Processing, Artificial Intelligence, Advanced Python; strong with PyTorch/TensorFlow, scikit-learn, Hugging Face ecosystem
Click on Apply to know more.