Founding AI Engineer (Agentic AI)

Executive Summary

The candidate Leković is a solid mid-to-senior ML/AI engineer with approximately 3 years of professional experience, currently working at a notable Serbian software company (VegaIT) while simultaneously contracted with a Palo Alto AI startup (ArchetypeAI). Their most relevant work — building an AWS Bedrock AgentCore multi-agent platform for fintech credit risk — maps closely to the agentic AI architecture demands of this role. They have demonstrated genuine technical leadership, production AI delivery, and multimodal ML breadth. The primary concerns are the absence of explicit LangChain ecosystem experience (LangGraph, LangSmith, etc.) and a lack of public engineering artifacts to validate code quality. Given the strong conceptual overlap between their AWS-native agentic work and the role's requirements, these gaps appear learnable. They are a credible FIT candidate who warrants an interview, with a structured technical assessment to verify engineering craft and ecosystem ramp-up ability.

Assessment Reasoning

The candidate meets the core minimum requirements of the role: 3+ years of ML/AI engineering experience, proven production AI systems delivery (multi-agent RAG platform in fintech), strong Python and cloud skills, and demonstrated ownership mentality including technical leadership of a 5-person team. Their AWS Bedrock AgentCore experience is functionally equivalent to the agentic AI patterns the role demands (multi-agent orchestration, RAG, tool calling, evaluation frameworks, observability, CI/CD). While they lack explicit experience with the LangChain ecosystem tools listed (LangGraph, LangSmith, LangFuse, LlamaIndex, CrewAI), these are adjacent to patterns they have clearly implemented via AWS-native services, and a competent senior engineer with this background should be able to ramp on these tools quickly. The absence of code samples and public GitHub is a real gap for a founding engineer hire and should be tested in the interview process. Overall, the candidate clears the 70+ threshold for FIT based on strong real-world agentic AI credentials, multimodal experience, and leadership trajectory, with confidence moderated by the missing ecosystem familiarity and lack of verifiable code artifacts.

Interview Focus Areas

No code example or GitHub profile was submitted, preventing direct assessment of code quality, style, or engineering depth. The resume narrative suggests solid system-level thinking and production engineering habits, but without artifacts to review, this dimension must be scored conservatively. This is a notable omission for a senior founding engineer candidate and should be addressed in the interview process with a take-home or live coding exercise.

The candidate is a capable ML/AI engineer with approximately 3 years of professional experience, including strong production credentials in agentic AI, RAG systems, and multimodal ML. Their AWS Bedrock AgentCore work is highly relevant to the role, demonstrating real ownership and delivery of complex AI products. The primary gap is familiarity with the specific LangChain ecosystem tooling (LangGraph, LangSmith, LangFuse, LlamaIndex) the job explicitly calls out, though their AWS-native equivalents suggest transferable patterns.