Experience: 2-3 years

About Riverfront
Riverfront is independent behavioral validation for AI agents. We test agents through their real interface, the way a user or regulator would, and produce evidence that boards, auditors, and regulators can rely on. Our methodology is aligned to the frameworks regulators are writing (EU AI Act, GDPR, UAE PDPL, Colorado AI Act, ISO/IEC 42001), and it's versioned end to end, so any result can be reproduced and defended months later.

We work in a spec-driven, AI-orchestrated model where the quality of your thinking matters more than your typing speed. We're looking for people who can understand ambiguous problems, make sound technical decisions, and build.

Location: Pune, India/Dubai, UAE

The Role
You will own complex problems from brief to delivery - understanding the problem, turning it into a accurate spec, making architectural decisions, building alongside the team, and shipping with confidence.
This is a hands-on engineering role with significant ownership.

What You'll Do




- Turn complex product and client requirements into structured technical specs before development begins
- Architect and build agentic AI systems end to end - including RAG pipelines, multi-agent orchestration, MCP-based integrations, tool-use workflows, and memory
- Make architecture decisions across model selection, retrieval strategy, memory design, and orchestration frameworks such as LangGraph, LangChain, Semantic Kernel, CrewAI, and AutoGen
- Own performance and cost trade-offs across token usage, latency, inference quality, and LLM providers including OpenAI, Anthropic, Vertex AI, Azure OpenAI, and AWS Bedrock
- Build observability and evaluation systems using tools such as Langfuse, LangSmith, Helicone, Arize, and W&B;
- Define evaluation criteria and determine what "good" actually means for an AI agent
- Contribute directly to Riverfront's AI governance and evaluation platform
- Participate in solution design, technical decisions, and product deli

📌 Ai Generalist (India)
🏢 Riverfront
📍 India