The Anatomy of a Self-Improving Agent w/ Arize

Part of The AI Builders Summit

Hosted by Kiriti & Aish and Aparna Dhinakaran

Wed, Sep 9, 2026

5:00 PM UTC (1 hour)

Virtual (Zoom)

Free to join

Invite your network

Go deeper with a course

Featured in Lenny’s List
Building Agentic AI Applications with a Problem-First Approach
Aishwarya Naresh Reganti and Kiriti Badam
View syllabus

What you'll learn

Instrument agents to surface failures

What to capture in traces and evals so a production failure is findable, not buried in dashboards

Move from dashboards to diagnosis

How observability is shifting from a human combing through traces to agent-driven investigation.

Close the loop from break to fix

The anatomy of a self-improvement loop: detect, diagnose, and propose a change.

Verify improvements before they ship

How evals turn a plausible change into a proven one, so an improvement does not regress once it is live.

Why this topic matters

As agents get more capable, the bottleneck shifts from building them to improving them. Observability and evals surface what an agent is doing, but a human still reads the dashboards and combs the traces to fix what broke. Aparna walks through how observability is moving from human-driven investigation to agent-driven execution, and the loop that turns a production break into a verified fix.

You'll learn from

Kiriti & Aish

AI practitioners with combined 20+ years of experience

Aishwarya Naresh Reganti is the founder and CEO of LevelUp Labs, a human-first AI startup that helps develop and deploy production-ready generative AI solutions for mid to large enterprise clients. With over 9 years of experience in machine learning, she has published more than 35 research papers at top-tier AI conferences, including NeurIPS, AAAI, and CVPR.

Kiriti Badam is a member of the technical staff at OpenAI Codex, with over a decade of experience designing high-impact enterprise AI systems. He specializes in AI-centric infrastructure, with deep expertise in large-scale compute, data engineering, and storage systems.

Aparna Dhinakaran

Co-Founder and Chief Product Officer at Arize AI

Aparna Dhinakaran is the Co-Founder and Chief Product Officer at Arize AI, a pioneer in AI observability and evaluation trusted by Fortune 100 companies. A frequent speaker at top conferences and a top-read technical voice in AI, her perspectives on AI engineering and agent infrastructure on X receive millions of views every month. Before Arize, she built for and led teams at Uber, Apple, and TubeMogul (acquired by Adobe), including Uber's core ML infrastructure platform, Michelangelo. She holds a B.S. in Electrical Engineering and Computer Science from UC Berkeley, where she published research with the Berkeley AI Research group.

See all products from Kiriti & Aish

Sign up to join this lesson

By continuing, you agree to Maven's Terms and Privacy Policy.