AI Architect

EdgeVerve Systems Limited; Texas; Ibrido; USD 180.000 - 240.000; Tempo pieno

Descrizione del lavoro

EdgeVerve Systems Limited is seeking an AI Architect to lead end-to-end GenAI and agentic AI solutions on the AI Next platform. You will decide model selections, design retrieval, orchestration, guardrails, and cost awareness, guiding client workshops to production.

You will mentor GenAI engineers and engineers in Python, cloud deployments (AWS/Azure/GCP), and scalable services, ensuring security, auditability, and reliable performance.

Competenze

• 10+ years in software or solution architecture across distributed, cloud-native or integration-heavy enterprise systems.
• 2+ years hands-on with GenAI, including architecting at least one LLM-based solution (RAG or Agentic).
• Agent design: tool calling, MCP, orchestration, state and memory, identity, authorization, error handling and human-in-the-loop.
• Understanding of end-to-end Intelligent Document Processing (IDP) solution architecture.
• Retrieval design: embeddings, vector and hybrid search, reranking, and measuring retrieval quality.
• Evaluation and monitoring: you have built evaluation sets and metrics, used them to decide releases, and monitored LLM systems in production.
• Practical GenAI security: defending against prompt injection, data leakage and over-privileged tools.
• Model trade-offs across at least two model families, including cost and latency.
• Strong Python; you can build a prototype to prove a design.
• Experience with cloud platforms (AWS, Azure, or GCP), Docker, Kubernetes, and CI/CD.

Mansioni

• Design end-to-end architecture with key decision and trade-offs for Generative AI and Agentic AI solutions on the AI Next platform, covering LLM selection, RAG pipelines, agent orchestration, tool integration, and guardrails.
• Architect agentic and multi-agent workflows using frameworks such as LangChain, LangGraph, and Google ADK.
• Design backend and integration layers using Python, exposing AI capabilities as secure and scalable services.
• Choose simplest design covering deterministic workflow, single agent or multiagent with focus on right autonomy level, security, MCP integrations, auditability and human review required for the client problem statement.
• Design enterprise retrieval with ingestion, chunking, hybrid search, reranking and permission-aware access to data sources with knowledge refresh strategies.
• Structured evaluation and recommendation of foundation models (for example Claude, Gemini, and OpenAI models) and deployment options for client problem statements.
• Design IDP (Intelligent Document Processing) solution architecture end to end, combining document ingestion, classification, OCR and layout-aware extraction, LLM-based extraction, validation, human-in-the-loop review, and downstream integration (good to have).
• Technical Leadership & Quality: Provide technical leadership and mentorship to Gen AI engineers, Python developers, and technology leads.

Strumenti

LangChain

LangGraph

Google ADK

Python

AWS

Azure

GCP

Docker

Kubernetes

SQL & Vector DBs

Descrizione del lavoro

Role Summary:

As AI Architect you are the design authority for Generative and Agentic AI solutions built on AI Next, from the first client workshop to production. You decide how a use case should be solved — model, retrieval, agent design, level of autonomy, guardrails — prove it with evaluations, and make sure it runs reliably, safely and at a known cost. The patterns you prove become AI Next reference architectures used by other teams.

Your success in a year looks like 2-3 GenAI or Agentic solution taken from design to production meeting the agreed outcome under the cost targets with cost per transaction and AI performance tracked.

Key Responsibilities

• Design end-to-end architecture with key decision and trade-offs for Generative AI and Agentic AI solutions on the AI Next platform, covering LLM selection, RAG pipelines, agent orchestration, tool integration, and guardrails.
• Architect agentic and multi-agent workflows using frameworks such as LangChain, LangGraph, and Google ADK.
• Design backend and integration layers using Python, exposing AI capabilities as secure and scalable services.
• Choose simplest design covering deterministic workflow, single agent or multiagent with focus on right autonomy level, security, MCP integrations, auditability and human review required for the client problem statement.
• Design enterprise retrieval with ingestion, chunking, hybrid search, reranking and permission-aware access to data sources with knowledge refresh strategies.
• Structured evaluation and recommendation of foundation models (for example Claude, Gemini, and OpenAI models) and deployment options for client problem statements.
• Design IDP (Intelligent Document Processing) solution architecture end to end, combining document ingestion, classification, OCR and layout-aware extraction, LLM-based extraction, validation, human-in-the-loop review, and downstream integration (good to have).
• Technical Leadership & Quality:
• Provide technical leadership and mentorship to Gen AI engineers, Python developers, and technology leads.
• Establish standards for prompt engineering, agent design, evaluation, and deployment of GenAI solutions.
• Conduct design and code reviews, and ensure reliability, security, scalability, and performance of AI solutions.
• Provide recommendations on optimizing cost per transaction adopting strategies like context management, routing, caching and batching.
• Troubleshoot and resolve technical issues in production AI systems.
• Evaluation, Governance & Responsible AI:
• Define tracing, observability, cost, and latency monitoring approaches for LLM and agent-based solutions.
• Define and set up evaluations to support solutions with success metrics agreed by client stakeholders, datasets, harness and gating for automated tests
• Define rollout, rollback and fallback strategies with owing technical incidents and escalations.
• Assess ethical implications, data privacy, and security risks of GenAI deployments, and design appropriate policies, controls, guardrails and human review where errors are costly.
• Design for auditability in the context of client’s regulatory context while collaborating with security and compliance teams.
• Be the primary technical contact for client architects and business stakeholders, drive GenAI architecture workshops and explain trade-offs in business terms.
• Turn GenAI and agentic AI use cases into value cases with expected benefits, cost, risks, and define path to production.
• Support pre-sales activities by providing GenAI solution design, estimates, and assumptions.
• Technical Leadership and Platform Impact
• Lead and mentor GenAI engineers, developers and technology leads enabling evaluation, design and operations.
• Turn proven patterns into AI Next reference architectures, templates and components.
• Assess new models, frameworks and standards against clear criteria before recommending adoption.
• Partner with product architecture, data, engineering, and security teams feeding them with gaps, client demands and failures thereby influencing the product roadmap.

Required Qualifications and Skills

• 10+ years in software or solution architecture across distributed, cloud-native or integration-heavy enterprise systems.
• 2+ years hands-on with GenAI, including architecting at least one LLM-based solution (RAG or Agentic).
• Agent design: tool calling, MCP, orchestration, state and memory, identity, authorization, error handling and human-in-the-loop, using a framework (e.g. LangGraph, Google ADK) or custom orchestration.
• Understanding of end-to-end Intelligent Document Processing (IDP) solution architecture, including document ingestion, extraction, validation, and integration.
• Retrieval design: embeddings, vector and hybrid search, reranking, and measuring retrieval quality.
• Evaluation and monitoring: you have built evaluation sets and metrics, used them to decide releases, and monitored LLM systems in production.
• Practical GenAI security: defending against prompt injection, data leakage and over-privileged tools.
• Model trade-offs across at least two model families, including cost and latency.
• Strong Python; you can build a prototype to prove a design.
• Experience with cloud platforms (AWS, Azure, or GCP), Docker, Kubernetes, and CI/CD.
• Strong understanding of data structures, databases (SQL and vector), and the AI project lifecycle.
• Understanding of AI security, data privacy, and responsible AI practices.
• Excellent communication skills and the confidence to lead client architecture discussions and explain trade-offs to non-technical client stakeholders.
• Good to have
• Hosting and optimizing open-weight models; knowing when fine-tuning or distillation beats prompting and retrieval.
• Designing and governing MCP servers or tool registries.
• AI governance frameworks: NIST AI RMF, ISO/IEC 42001, EU AI Act.
• Cloud or cloud-AI certifications.
• Standardized tracing for LLM and agent systems (e.g. OpenTelemetry GenAI).
• Agent-to-agent interoperability (e.g. A2A) across frameworks or vendors.
• Open-source contributions, publications or talks on applied GenAI.