
Sign up to save your podcasts
Or


Kiriti was an early member of the Codex team at OpenAI and is now CTO at Level Up Labs. We go deep on the engineering behind coding agents: what a "harness" actually is and why it matters as much as the model, how compaction and prompt caching work under the hood (including the bug that paged the Codex team every night), how you evaluate an agent when the model keeps changing, and how he personally ships large features with parallel agents and automated code review.
If you're into AI engineering, harness engineering, prompt engineering, or loop engineering, this one's for you.
Chapters
0:00 Intro
0:32 Meet Kiriti
1:26 Prompt vs. harness
4:05 Inside a harness
6:21 Who needs harnesses
7:40 Tool search
8:40 Better harness wins
11:49 How compaction works
14:28 Server-side compaction
15:49 Evaluating harnesses
18:21 Prompt caching
21:56 Front-end taste
24:58 Monitoring live sessions
27:05 Learning from traces
30:23 Dropping in GPT-7
32:27 Leaving OpenAI
35:01 Startup vs. lab
36:32 Right-sizing models
38:42 Handwritten code?
39:47 His feature workflow
44:10 His agent setup
45:50 Code review now
48:55 Reviewing at speed
50:08 Getting good
53:12 Find Kiriti
Links
Kiriti on LinkedIn: https://www.linkedin.com/in/sai-kiriti-badam/
Kiriti's Substack: https://substack.com/@kiritib
Kiriti's Maven course: https://maven.com/aishwarya-kiriti/genai-system-design
Free lessons on Maven: https://maven.com/search?query=kiriti+badam&tab=free-lessons
Level Up Labs: https://levelup-labs.ai/
Sam Hall is the founder of Wafer Systems. He skipped college, taught himself how to build, landed engineering jobs through cold outreach, and eventually raised $1.5M to start his own company.
We talk about breaking into software without traditional credentials, why coding interviews don’t always reflect real engineering ability, what it’s actually like to raise venture capital, and the technical challenges behind building Wafer.
Follow Sam on X: https://x.com/Shmall
Check out Wafer: https://wafer.systems/
Follow Wafer on X: https://x.com/WaferSystems
1:02 Meet Sam Hall
1:33 Growing up curious and discovering coding
4:00 Building rockets, drones, and an early AI project
6:51 The unpaid labor
10:43 First business
14:07 Skipping college
19:06 Cold outreach
20:09 Early startup life
22:26 Moving to California
29:09 Deciding to build a company
34:21 Building an AI-native phone from scratch
37:30 Raising a pre-seed round
39:17 Finding community
41:32 Product demo
45:09 Tech deep dive
51:26 Building through trial and error
52:36 Retrieval systems, search, and AI context
59:52 Where to find Sam and Wafer
We get into why you can’t fully trust an agent, sandboxing and credential isolation, when agents are better than traditional software, the security problem that led Gavriel to build NanoClaw, why NanoClaw is dramatically smaller than OpenClaw, and how his team uses agents to test and review their own code.
If you're building with Claude Code, Codex, OpenClaw, or autonomous agents in general, this one goes deep on where agentic software is headed.
Timestamps
0:00 Intro
1:43 Agent security & why you can’t trust an agent
8:19 Will AI models become secure enough to trust?
11:15 Why long-running agents go off track
21:32 When to use AI agents vs. traditional software
23:25 How NanoClaw started
30:20 NanoClaw vs. OpenClaw: security, vibe coding & code size
42:49 What makes NanoClaw different
46:00 Why taste can matter more than evals
49:09 Using AI agents to review and test code
Aishwarya Reganti joins me to break down how to become an AI engineer in 2026, including the skills software engineers should learn, how AI engineering interviews are changing, and what companies actually look for when hiring. We cover system design, evals, context engineering, AI agents, production reliability, AI product sense, and why judgment matters more than simply generating code with AI. If you’re a software engineer looking to break into AI engineering, this is for you.
Aishwarya's GitHub: https://github.com/aishwaryanr/awesome-generative-ai-guide
Aishwarya's LinkedIn: https://www.linkedin.com/in/areganti/
Learn from Aishwarya directly on Maven: https://maven.com/aishwarya-kiriti/genai-system-design
0:00 Intro
2:21 What is AI Engineering?
7:51 Evals
15:05 Interviewing for AI Eng and FDE roles
18:58 AI Product Sense
23:01 Enterprise AI Use Cases & Context Engineering
32:19 When to Use AI and Why AI Systems Fail in Production
42:08 Multi-Agent Systems and Coding Agent Workflows
48:26 Breaking into AI Eng as a SWE
Brandon & Arjay go into tactical details on exactly what it takes to land your first and second promotions as a software engineer in big tech
Shreyans Bhansali was the first engineer at Venmo, the founder of Socratic, and later joined Google when Socratic was acquired, then was a co-founder of Maven.
In this episode, we talk about what Venmo was actually like in the early days, we get into the story behind Socratic, how the Google acquisition happened, and what changed after moving from a startup into Big Tech. Later, we talk about AI, software engineering, and why Shreyans thinks the current AI moment feels similar to the early 2010s startup wave: obvious in hindsight, but still full of unsolved problems.
Sam Gardner recently joined the go-to-market team at Anthropic after his previous company, Stainless, was acquired by Anthropic.
Sam is not a software engineer, but he is one of the most technical non-engineers I’ve talked to. He represents a new kind of operator who sits between sales, product, engineering, and automation, and uses code and AI to stitch all of it together.
In this episode, we talk about what go-to-market actually looks like at a company like Anthropic, how Sam uses AI in his day-to-day work, how he broke into companies like Ramp, Salesforce, Stainless, and Anthropic, and where AI is useful versus where it becomes a crutch.
If you’re interested in software engineering, AI, startups, or the future of work, I think you’ll get a lot out of this conversation.
This is a conversation with Jeff Escalante, who’s currently a Director of Engineering at Clerk, where he spends more than 80% of his time writing code.
Before that, Jeff was an early engineer at HashiCorp, where he joined around 200 employees and stayed through the company growing to thousands of employees and eventually IPOing. During that time, he also became a prominent contributor to Next.js, which led him to join Vercel as the manager of the Next.js team, where he helped shape the roadmap for one of the most important frameworks in web development.
In this episode, we talk about how Jeff broke into software engineering without a traditional CS background, how to get hired without spraying your résumé everywhere, what he’s learned from spending most of his career in dev tools, how Clerk thinks about engineering, and how AI is changing the way software engineering actually works.
Hope you enjoy the first episode, and welcome to the podcast.