Technical Architect - Data Science & Gen AI

Job description
Primary Skills (Must Have experience):

• Generative AI Solution Architecture (2-3 years): Proven experience in designing and architecting GenAI applications, including Retrieval-Augmented Generation (RAG), LLM orchestration (LangChain, LangGraph), and advanced prompt design strategies.
• Backend & Integration Expertise (5+ years): Strong background in architecting Python-based microservices, APIs, and orchestration layers that enable tool invocation, context management, and task decomposition across cloud-native environments (Azure Functions, GCP Cloud Functions, Kubernetes).
• Enterprise LLM Architecture (2-3 years): Hands-on experience in architecting end-to-end LLM solutions using Azure OpenAI, Azure AI Studio, Hugging Face models, and GCP Vertex AI, ensuring scalability, security, and performance.
• RAG & Data Pipeline Design (2-3 years): Expertise in designing and optimizing RAG pipelines, including enterprise data ingestion, embedding generation, and vector search using Azure Cognitive Search, Pinecone, Weaviate, FAISS, or GCP Vertex AI Matching Engine.
• LLM Optimization & Adaptation (2-3 years): Experience in implementing fine-tuning and parameter-efficient tuning approaches (LoRA, QLoRA, PEFT) and integrating memory modules (long-term, short-term, episodic) to enhance agent intelligence.
• Multi-Agent Orchestration (2-3 years): Skilled in designing multi-agent frameworks and orchestration pipelines with LangChain, AutoGen, or DSPy, enabling goal-driven planning, task decomposition, and tool/API invocation.
• Performance Engineering (2 - 3 years): Experience in optimizing GCP Vertex AI models for latency, throughput, and scalability in enterprise-grade deployments.
• AI Application Integration (2 - 3 years): Proven ability to integrate OpenAI and third-party models into enterprise applications via APIs and custom connectors (MuleSoft, Apigee, Azure APIM).
• Governance & Guardrails (1 - 2 years): Hands-on experience in implementing security, compliance, and governance frameworks for LLM-based applications, including content moderation, data protection, and responsible AI guardrails.

Role & responsibilities :
As a Technical Architect specializing in LLMs and Agentic AI, you will own the architecture, strategy, and delivery of enterprise-grade AI solutions. You will work with cross-functional teams and customers to define the AI roadmap, design scalable solutions, and ensure responsible deployment of Generative AI across the organization:

Responsibilities:

• Architect Scalable GenAI Solutions: Lead the design of enterprise architectures for LLM and multi-agent systems, ensuring scalability, resilience, and security across Azure and GCP platforms.
• Technology Strategy & Guidance: Provide strategic technical leadership to customers and internal teams, aligning GenAI projects with business outcomes.
• LLM & RAG Applications: Architect and guide development of LLM-powered applications, assistants, and RAG pipelines for structured and unstructured data.
• Agentic AI Frameworks: Define and implement agentic AI architectures leveraging frameworks like LangGraph, AutoGen, DSPy, and cloud-native orchestration tools.
• Integration & APIs: Oversee integration of OpenAI, Azure OpenAI, and GCP Vertex AI models into enterprise systems, including MuleSoft Apigee connectors.
• LLMOps & Governance: Establish LLMOps practices (CI/CD, monitoring, optimization, cost control) and enforce responsible AI guardrails (bias detection, prompt injection protection, hallucination reduction).
• Enterprise Governance: Lead architecture reviews, governance boards, and technical design authority for all LLM initiatives.
• Collaboration: Partner with data scientists, engineers, and business teams to translate use cases into scalable, secure solutions.
• Documentation & Standards: Define and maintain best practices, playbooks, and technical documentation for enterprise adoption.
• Monitoring & Observability: Guide implementation of AgentOps dashboards for usage, adoption, ingestion health, and platform performance visibility.