AI Architech

Meeru AI Inc•Pleasanton, CA

20d

About The Position

Meeru AI is building an AI-native platform for finance and accounting teams. The platform connects to enterprise financial systems, including ERPs, CRMs, billing platforms, and HRIS, and uses machine learning to transform fragmented operational data into grounded, auditable intelligence for CFOs, controllers, and FP&A leaders. It supports SaaS multi-tenant, SaaS single-tenant, and on-premises deployments across AWS, Azure, and GCP, serving Fortune 500 finance teams with data-isolation, auditability, and compliance needs. The AI Architect will help design and build the platform’s intelligence engine and conversational AI agents. The engine computes reproducible, traceable answers from verified source data; agents interpret user requests, route them to the engine, and present results in natural language without inventing answer content. This is a hands-on architecture role that sets technical direction for the engine and agent layer, partners with US-based AI leadership and the Staff AI Engineer on grounding and evaluation rigor, and promotes deterministic, auditable AI practices across the team. The role is based in Hyderabad (preferred) or can be remote elsewhere, with a few hours of daily overlap with US time zones. Engagement may be full-time, contract, or contract-to-hire.

Requirements

Senior- or staff-level applied AI/ML experience.
Experience shipping production AI systems and leading the technical direction of at least one system.
Hands-on fluency with modern LLM patterns, including retrieval-augmented generation (RAG), tool-using agents, and structured output such as JSON schemas or function calling.
Experience building multi-step agent workflows with LangGraph or a similar framework, including verification steps, fallbacks, and human confirmation points.
Ability to design an engine, its data interfaces, and the agent layer as one coherent, maintainable system, and document decisions such as architecture decision records.
Experience enforcing a no-unsupported-answers rule in a real system and measuring faithfulness and hallucination with automated evaluations, ideally running in CI.
Knowledge of techniques for making the same question against the same data return the same answer on top of probabilistic models.
Sound judgment about what a model can legitimately claim, particularly when answers must be computed and traceable in a high-stakes domain.
Production software engineering experience in Python, including writing robust, tested services.
SQL access to warehouses or databases such as Snowflake or PostgreSQL, and comfort with containerized deployment.
Working knowledge of vector stores such as pgvector, Pinecone, Weaviate, or Qdrant, and of computing reliable answers over structured data.
Technical leadership skills, including mentoring engineers, running design reviews, and translating strategy into execution.
Strong English communication skills for explaining architecture trade-offs to technical and non-technical stakeholders.
Based in Hyderabad (preferred) or remote elsewhere, and able to keep a few hours of daily overlap with US time zones.

Nice To Haves

Experience in finance, accounting, or another audited domain such as healthcare, insurance, or legal.
Experience with knowledge graphs, semantic layers, or rule/reasoning engines that model facts and relationships.
Experience designing human-in-the-loop review, confirmation, or feedback loops that improve a system over time.
Hands-on experience with managed LLMs inside customer clouds, such as Amazon Bedrock, Google Vertex AI, or Azure OpenAI deployed in a customer’s private network or VPC.
Experience shipping AI to AWS, Azure, and GCP, including single-tenant and on-premises installations, and working with platform teams on model serving and inference cost.
Experience with tracing and evaluation tooling for LLM applications.

Responsibilities

Design and build the platform’s intelligence engine and conversational AI agent layer.
Set technical direction for the engine and agent layer, designing them as a coherent, maintainable system with appropriate data interfaces.
Build systems that produce deterministic, reproducible, explainable answers grounded in verified source data and traceable to the source data and logic used.
Develop conversational agents that interpret requests, route them to the engine, and present results without fabricating answer content.
Partner with US-based AI leadership and the Staff AI Engineer on grounding and evaluation rigor.
Establish and improve practices for deterministic, auditable AI across the team.
Design and document architecture decisions, including architecture decision records.
Mentor engineers, run design reviews, and turn strategy set by US leadership into well-built execution.
Explain architecture trade-offs to engineers, product managers, and non-technical stakeholders.
Maintain a few hours of daily overlap with US time zones.