The Role

Doers first. Leaders by example. This is not a management role.

The Senior Consultant — AI & Agentic Solutions at Aarvian is the technical backbone of a cross-functional AI product team working on real, at-scale enterprise AI problems. You will spend 80% of your time in the product — designing, architecting, and shipping production Agentic AI solutions deployed in live business environments on Azure and Databricks. The remaining 20% you act as the quality gate for the team: reviewing architectures, guiding technical approaches, and holding the standard of everything that ships.

We are not building chatbots or dashboards. We build AI products that execute autonomous business actions at scale.

You will work directly alongside the AI/ML CoE owner — getting unfiltered exposure to how a high-growth AI adoption startup identifies, scopes, and delivers use cases that move commercial metrics for retail clients.

What You Will Do

Hands-On Development — 80%

- Design and build production Agentic AI products end-to-end — from multi-agent orchestration and RAG pipelines to LLM-powered automation workflows. Not POCs. Not notebooks. Shipped, monitored, production systems.
- Multimodal AI development: design pipelines that process text, images, and video streams using Azure OpenAI (GPT-4o, GPT-4V), Azure AI Vision, and Azure Video Indexer — integrated into scalable backend services.
- Content Generation System: build robust generation pipelines with prompt engineering, output validation, JSON/structured extraction, and quality-gate loops using LangChain, Semantic Kernel, or LlamaIndex.
- Real-time and batch AI pipelines: architect streaming data flows on Azure Event Hubs and Databricks Structured Streaming; integrate AI inference with Azure Functions, Azure Container Apps, and AKS.
- Agentic workflow orchestration: build multi-agent systems using AutoGen, LangGraph, CrewAI, or Semantic Kernel with tool use, agent memory, state management, and error recovery at enterprise scale.

Technical Leadership — 20%

- Quality gate: every significant AI product architecture, agent design, and prompt engineering approach produced by the team goes through your review before it reaches the client or production.
- Solution design: frame the technical approach for recent AI product use cases — selecting the right architecture, identifying data and integration requirements, and producing design documents the team can build from.
- Mentoring by doing: pair on hard problems, review pull requests with substantive feedback, and raise the technical bar by demonstrating it — not by describing it.
- Process-first discipline: you insist on understanding the business workflow and data realities before selecting an AI approach.
- Working with the CoE: you are the primary technical interface between the delivery team and the AI CoE lead — translating business intent into architecture and back again.

What We Are Looking For Must-Have Experience





- 6+ years of professional experience building and shipping AI/Gen-AI/Agentic AI products in enterprise environments — with a career that has stayed close to the code throughout. Not years of managing teams who build products.
- Agentic AI depth: at least one production or near-production Agentic AI system — a multi-agent workflow executing a business process end-to-end, a RAG pipeline surfacing decision-relevant information, or an LLM-orchestrated solution integrated with live operational systems. At enterprise scale.
- Agentic frameworks: LangChain, LangGraph, LlamaIndex, AutoGen, CrewAI, or Semantic Kernel. Production-grade: tool use, memory, orchestration patterns, failure modes at scale.
- Azure AI Platform: Azure OpenAI Service (GPT-4o, GPT-4V, Whisper, DALL-E), Azure AI Studio, Azure AI Search (vector + hybrid search), Azure Cognitive Services, Azure Vision, Azure Video Indexer.
- Databricks: Delta Lake, Unity Catalog, Databricks Model Serving, MLflow (for tracking AI experiments and model versions), Databricks Workflows for pipeline orchestration. Comfortable with PySpark for large-scale data transformation feeding AI pipelines.
- LLMOps & AI observability: PromptFlow, Langfuse, Azure AI Studio evals, or equivalent. Cost tracking, latency monitoring, hallucination and quality measurement in production.
- Cloud-native deployment: Azure Container Apps, AKS, Azure Functions for AI inference endpoints; Azure Event Hubs and Databricks Structured Streaming for real-time AI pipelines; Azure Service Bus for async agent communication.
- Vector Database: Azure AI Search, Qdrant, Weaviate, or Pinecone — index design, embedding strategy, retrieval optimisation, hybrid search.

Agentic AI — Depth, Not Surface

- This is not a chatbot role. We are looking for someone who has built Agentic AI systems that execute autonomous business actions — not conversational UIs, not FAQ bots, not demo-ware. If your AI experience is primarily chatbot or virtual assistant development, this is not the right environment.
- Real Agentic AI exposure: at least one production or near-production system — a multi-agent workflow that executes a business process end-to-end, a RAG pipeline that surfaces decision-relevant information from complex enterprise data, or an LLM-orchestrated solution that integrates with live operational systems. At industrial or enterprise scale. A tutorial project does not count.
- Python hands-on: expert. Modular, testable, production-quality code. FastAPI for AI service layers, async patterns, structured output parsing. You would not be embarrassed to have a senior engineer review it.




- Retail or consumer domain (good to have): experience with e-commerce, CPG, FMCG, or equivalent consumer-facing industries. Strong candidates from other enterprise verticals with equivalent AI product depth will also be considered.

The Startup Dimension

- Pace: you are comfortable with ambiguity, incomplete requirements, and shifting priorities. A startup does not have the luxury of perfect problem statements.
- Ownership: you treat the client's problem as your own. You do not wait to be told what to do when something is broken.
- Communication: you can explain a model, a risk, or a trade-off to a business stakeholder without losing them or condescending to them.
- 5 days in-office, Bengaluru: this role requires you to be present. The team is co-located by design.

What We Are Not Looking For We will not compromise on three things.

First, hands-on depth: if your last 12 months has been primarily stakeholder management, dashboards, documentation for other people's work, or team management without personally writing production code — this is not the right fit.

Second, real scale: if "deployed" in your experience means a pilot that ran for 6 weeks on a sample dataset, that is not what we mean. We mean models running today, at volume, in a live business environment.

Third, AI substance: if your AI experience is primarily chatbot or virtual assistant development, we are building something materially different. We work on forecasting, optimisation, Agentic workflows, and computer vision — systems that make or influence business decisions at scale.

Senior at Aarvian means technically senior, not organisationally senior.

Who You Are

Builder — You find more satisfaction in a model that ships to production than in one that scores well in a notebook. Building is how you think.

Commercially Grounded — You frame model quality in margin recovered, churn reduced, or waste eliminated — not just in AUC or RMSE. Metrics exist to serve the business.

Technically Uncompromising — You will push back on a bad architecture decision. Technical debt is not a strategy you quietly accept to meet a deadline.

Startup-Wired — You are energised by fast cycles, real ownership, and the chance to build something from the ground up — not frustrated by the absence of process.

Why Aarvian

- Real retail problems, real production. The models you build at Aarvian affect actual buying decisions, pricing calls, and inventory levels at live retailers — not internal tooling.
- Direct CoE access. You work shoulder-to-shoulder with the AI/ML CoE owner — not filtered through three layers of management. Your voice shapes technical direction.
- Startup exposure. You will see — and participate in — how a high-growth AI adoption business is built: how use cases are sold, scoped, delivered, and scaled.
- Genuinely interesting technical work. Advanced ML, Agentic AI, retail domain complexity. Not the same use case recycled for different logos.

📌 Senior Consultant - AI Engineer (Bengaluru)
🏢 Aarvian
📍 Bengaluru