About the roleRiyaLabs is building enterprise AI software that enables organizations to deploy AI coworkers and coordinated AI teams for real business workflows.

Our core platform, RiyaLabs Studio, connects AI coworkers to enterprise knowledge, data, and approved tools. RiyaLabs Trust Gateway provides the governance and controlled-access foundation around model interactions and tool execution.

We are seeking a hands-on AI Architect to design and deliver secure, scalable, and maintainable AI systems spanning agentic workflows, advanced retrieval, natural-language data interaction, and voice-based experiences. This is a product-engineering role involving architecture ownership, implementation, code reviews, and technical leadership—not a purely advisory position.

The role supports enterprise workflows across finance, HR, procurement, business intelligence, operations, customer service, and cybersecurity.

Key responsibilities:Architecture and technical leadership

- Own end-to-end AI architecture across data ingestion, retrieval, orchestration, model access, tool execution, backend services, and user experience.
- Translate business requirements into technical designs, delivery milestones, and measurable acceptance criteria.
- Define explicit boundaries between AI reasoning, deterministic business logic, policy enforcement, and human approval.
- Evaluate build-versus-buy decisions and document trade-offs involving reliability, security, latency, cost, and maintainability.
- Develop reference implementations, review code, resolve technical blockers, and mentor engineers.
- Maintain architecture diagrams, decision records, integration contracts, and deployment documentation.

Agentic AI and enterprise integration
- Design single-agent and multi-agent workflows with defined responsibilities, shared context, state management, and escalation paths.
- Architect MCP servers and clients, enterprise API integrations, and event-driven workflows.
- Implement controlled tool execution with validated inputs, structured outputs, retries, timeouts, idempotency, and recovery.
- Connect AI coworkers to document repositories, databases, CRM, ERP, ITSM, and other approved enterprise systems.
- Build human-in-the-loop workflows for consequential actions and ambiguous decisions.

Advanced RAG and enterprise knowledge
- Architect advanced RAG pipelines covering document extraction, OCR, section-aware chunking, metadata enrichment, indexing, and incremental synchronization.
- Combine keyword search, vector retrieval, metadata filtering, and reranking based on the use case.
- Design query rewriting, query decomposition, multi-step retrieval, and contextual retrieval workflows.
- Evaluate agentic retrieval, graph-based retrieval, and other advanced approaches where they offer measurable value.
- Preserve document permissions, source provenance, version history, and evidence links throughout retrieval and generation.
- Design retrieval across both structured and unstructured enterprise information.




- Establish evaluation datasets and metrics for retrieval relevance, coverage, grounding, answer quality, and appropriate abstention.
- Address stale content, conflicting evidence, duplicate documents, extraction failures, and unsupported responses.

Text-to-SQL and semantic data agents
- Architect natural-language-to-SQL workflows that translate business questions into controlled database queries.
- Design semantic layers incorporating business definitions, metrics, relationships, approved joins, and schema context.
- Implement schema retrieval, SQL generation, validation, execution controls, and evidence-based result interpretation.
- Enforce read-only access where appropriate, row-level restrictions, query limits, timeouts, and sensitive-data controls.
- Handle ambiguous questions, incorrect joins, unsupported metrics, and requests requiring clarification.
- Build workflows that turn query results into explanations, visualizations, dashboards, and actionable insights.
- Evaluate SQL correctness, execution success, business-semantic accuracy, and result consistency.
- Support relational databases and enterprise data platforms, with specific technology choices driven by product requirements.

Voice agents and conversational systems
- Architect voice agents combining speech-to-text, conversational orchestration, tool execution, and text-to-speech.
- Design streaming interactions with low-latency responses, interruption handling, turn detection, and conversational state.
- Build multilingual experiences and evaluate recognition quality across accents, noise conditions, and domain terminology.
- Integrate voice agents with telephony, web applications, contact-centre systems, and approved enterprise tools.
- Implement identity verification, consent workflows, controlled actions, and human handoff where required.
- Address transcription errors, dropped connections, repeated requests, and incomplete conversations.
- Evaluate end-to-end latency, task completion, transcription quality, tool-use reliability, and handoff effectiveness.
- Define secure handling and retention of audio, transcripts, and associated customer information.

Security, governance, and production delivery
- Design identity-aware access, tenant isolation, least-privilege permissions, and protected credential handling.
- Implement policy-controlled model and tool access, approval-bound execution, and auditable workflow records.
- Address prompt injection, sensitive-data exposure, unsafe tool use, and cross-tenant access risks.
- Define data-processing boundaries and deployment options for cloud and customer-controlled environments.




- Establish containerized deployment, CI/CD, automated testing, monitoring, and incident-response readiness.
- Track workflow execution, model usage, latency, errors, cost, and operational outcomes.
- Build regression tests and release gates for retrieval, SQL generation, agent behavior, and voice workflows.

Required qualifications
- Strong software-engineering foundation and experience architecting production applications.
- Hands-on Python proficiency and experience building backend services, APIs, and asynchronous workflows.
- Demonstrated delivery of LLM-powered systems beyond prototypes.
- Practical experience with advanced RAG, including hybrid retrieval, reranking, metadata filtering, and source-grounded responses.
- Experience building Text-to-SQL or comparable natural-language data-query systems, supported by strong SQL and data-modelling skills.
- Hands-on experience with voice agents or real-time conversational AI, including speech services and enterprise integrations.
- Experience designing agent workflows, tool integrations, or comparable orchestration systems.
- Knowledge of distributed systems, queues, caching, authentication, authorization, and resilient service design.
- Experience with Docker, CI/CD, and at least one major cloud platform.
- Ability to evaluate architecture choices through testing and communicate trade-offs clearly.
- Evidence of technical ownership, implementation, design reviews, and team mentoring.

Depth in one or more of advanced RAG, Text-to-SQL, and voice agents is essential, alongside the ability to architect and guide delivery across all three. Preferred experience
- MCP implementation, agent frameworks, and durable workflow engines.
- Graph-based retrieval, multimodal retrieval, and document-understanding pipelines.
- Semantic layers, enterprise analytics, and platforms such as Snowflake.
- Telephony integration, WebRTC, SIP, or contact-centre workflows.
- Multi-tenant SaaS architecture, Kubernetes, and infrastructure as code.
- AI gateways, model routing, inference optimization, and open-weight model deployment.
- Enterprise identity, policy engines, approval workflows, and observability.
- Banking, financial services, cybersecurity, or other regulated environments.

We value demonstrated delivery and sound architectural judgment over familiarity with every listed framework or platform.

Application detailsPlease share

- Your updated resume.
- A brief description of one or two production AI systems you have architected or built, explaining your contribution.
- Relevant examples involving advanced RAG, Text-to-SQL, voice agents, or agentic workflows.
- GitHub repositories, portfolio links, or nonconfidential technical work samples, if available.
- Your current location and willingness to work in Jaipur.
- Your notice period and earliest joining availability.
- Your compensation expectations.

Please do not share confidential code, customer data, or proprietary documents from current or previous employers.

📌 AI Architect (India)
🏢 riyalabs.ai
📍 India