Coding Chats

Coding Chats Episode 80 - start-up advisor Alexander Berkovich shares his expertise on building successful start-ups, hiring strategies, CTO roles, and the importance of communication between technical and business teams. Discover practical tips for navigating the challenges of early-stage companies and how to align technical excellence with business goals.Chapters00:00 Introduction to Start-up Advising02:09 The Day-to-Day of a Start-up Advisor05:39 Hiring Challenges in Start-ups07:39 Defining the Role of a CTO10:36 Common Mistakes in CTO Hiring12:59 Bridging the Gap: Technical and Business Communication16:40 Utilizing Client Feedback for Product Improvement20:06 Transitioning from Proof of Concept to Product24:01 Exploring Computer Vision in AI24:06 Balancing Technical Excellence and Business Focus24:09 Exploring Related ContentAlex's Links:Alex's LinkedIn: https://www.linkedin.com/in/alexander-berkovich-startup-advisor/John's Links:John's LinkedIn: https://www.linkedin.com/in/johncrickett/John’s YouTube: https://www.youtube.com/@johncrickettJohn's Twitter: https://x.com/johncrickettJohn's Bluesky: https://bsky.app/profile/johncrickett.bsky.socialCheck out John's software engineering related newsletters: Coding Challenges: https://codingchallenges.substack.com/ which shares real-world project ideas that you can use to level up your coding skills.Developing Skills: https://read.developingskills.fyi/ covering everything from system design to soft skills, helping them progress their career from junior to staff+ or for those that want onto a management track.TakeawaysA CTO's value is in leadership and strategy, not just how much they can code.AI has fundamentally changed the hiring process and what makes a good candidate.Document every corner you cut in a POC — it will catch up with you later.Engineers should seek direct client feedback to understand the real impact of their work.Communication is the most underrated skill in any startup team.A POC and a product are very different things — don't let one accidentally become the other.Startups offer breadth of experience that large enterprises simply can't match.Hiring for the right mindset matters more than hiring for pure technical skill.Small technical decisions can ripple out and affect cost, timelines, and the whole product.The best teams stay connected to the end goal, not just the task in front of them.

Coding Chats Episode 79 - Richmond Alake, Director of AI Developer Experience at Oracle, joins John to discuss agent memory — how AI agents store, retrieve, and adapt to information. He argues that developers building memory on flat files are naively reinventing the database, and that once you factor in concurrency, security, and scalability, a proper database is inevitable. The conversation covers the full memory stack and how Oracle's AI database keeps embeddings and data together without shipping sensitive information to external providers.The pair also explore why memory is the most universally relatable concept in AI, the history of how neuroscience shaped LLMs, and the problem of Catastrophic Forgetting that still haunts models today. A sharp AGI debate lands on a sobering point: an LLM is just a function — tokens in, tokens out — and most AI engineers are unknowingly rediscovering solutions that database engineers spent decades building.Chapters00:00 — What Is Agent Memory and How Does It Work?05:00 — File System vs Database: Which Should You Use for Agent Memory?09:00 — Why Building on Files Means You'll Reinvent the Database13:00 — How Oracle Is Meeting AI Developers Where They Are15:00 — Why Memory Is the Most Universal Concept in AI21:00 — From Computer Vision to LLMs: How Richmond Found His Path24:00 — Catastrophic Forgetting: The Problem That Hasn't Gone Away26:00 — Is AGI Real? Why the Goalposts Keep Moving33:00 — Handling PII, Data Sovereignty, and Access Control in AI Apps42:00 — The Rise of Memory Engineering: AI's Most Underrated DisciplineRichmond's Links:LinkedIn: https://www.linkedin.com/in/richmondalake/X: https://x.com/richmondalake John's Links:John's LinkedIn: https://www.linkedin.com/in/johncrickett/John’s YouTube: https://www.youtube.com/@johncrickettJohn's Twitter: https://x.com/johncrickettJohn's Bluesky: https://bsky.app/profile/johncrickett.bsky.socialCheck out John's software engineering related newsletters: Coding Challenges: https://codingchallenges.substack.com/ which shares real-world project ideas that you can use to level up your coding skills.Developing Skills: https://read.developingskills.fyi/ covering everything from system design to soft skills, helping them progress their career from junior to staff+ or for those that want onto a management track.Takeaways:File systems are fine for prototyping, but the moment you hit production scale you're just slowly reinventing the database.File systems are fine for prototyping, but the moment you hit production scale you're just slowly reinventing the database.Agent memory isn't a new concept — it's data management, and database engineers have been solving it for decades.Memory is the single most relatable entry point for explaining AI to anyone, technical or not.Catastrophic Forgetting isn't a solved problem — it plagued RNNs and still quietly haunts LLMs today.An LLM is ultimately just a function: tokens in, tokens out — which should temper any claims about sentience or AGI.The definition of AGI keeps shifting to match whatever AI can't do yet, making the whole debate almost meaningless.Most AI engineers have less than ten years of experience and are unknowingly rediscovering solutions that search and database engineers spent decades building."Vector search is all you need" is one of the most dangerous oversimplifications in AI engineering right now.Memory engineering — the crossover between data engineering, search optimisation, and agent design — is an emerging discipline that doesn't have a name yet but absolutely should.The real moat in AI products isn't the LLM itself, it's everything built around it — the harness, the memory, the retrieval pipeline.

Coding Chats Episode 78 - John Crickett talks to Robert Harris, an experienced engineering leader. Robert shares hard-won lessons from years of leading software teams, drawing on a distinctive "human systems" lens to explain why so many engineering organisations struggle — not because of bad people, but because of broken systems, misaligned leadership, and invisible cultural forces.The conversation weaves together philosophy, practical management advice, and candid personal anecdotes, making it equally relevant for first-time engineering managers and seasoned CTOs. The central thread throughout is that software is fundamentally a human endeavour, and leaders who treat it like a purely technical one will keep running into the same problems.Chapters0:00 — Every Problem is a Systems Problem3:00 — Labelling vs. Diagnosing: The Human Systems Approach6:15 — Poor Performance Is a System Failure, Not a People Failure9:10 — AI, Flat Orgs, and the Pressure on Engineering Managers11:30 — Diagnosing a Broken Team: A Real-World Turnaround24:05 — People Are Not Interchangeable Components26:00 — Culture: What Happens When Nobody's Watching33:00 — The Power Gradient and Cross-Team Collaboration39:00 — The C-Suite Distance Problem42:00 — Building Culture in Remote and Distributed Teams46:00 — Software Engineering Is a HumanityRobert's Links:https://www.linkedin.com/in/robert-n-harris/coded2lead.comJohn's Links:John's LinkedIn: https://www.linkedin.com/in/johncrickett/John’s YouTube: https://www.youtube.com/@johncrickettJohn's Twitter: https://x.com/johncrickettJohn's Bluesky: https://bsky.app/profile/johncrickett.bsky.socialCheck out John's software engineering related newsletters: Coding Challenges: https://codingchallenges.substack.com/ which shares real-world project ideas that you can use to level up your coding skills.Developing Skills: https://read.developingskills.fyi/ covering everything from system design to soft skills, helping them progress their career from junior to staff+ or for those that want onto a management track.TakeawaysPeople run on emotion and safety, not logic — lead them accordingly.When someone underperforms, look at the system before you look at the person.Labelling people as "difficult" or "lazy" is a way of avoiding the real problem.AI is accelerating code generation, but the human bottleneck downstream is getting worse, not better.The institutional memory inside a team is worth far more than anything in your wiki.Culture is what happens when nobody's watching — not what's written on the wall.If you send Slack messages at 10pm, your team will think there's no such thing as work-life balance.Only authorised people should authorise work — casual remarks from leaders land as commands.Co-location without connection isn't culture, it's a terrarium.Computers are a science, but software is a humanity.

Coding Chats Episode 77 — Arnaud Fournier, Forward Deployed Engineer at OpenAI, talks to John Crickett about how AI is fundamentally reshaping software engineering. He explores how OpenAI's own engineers have largely moved away from writing code line-by-line, shifting instead to what he calls "harness engineering" — orchestrating agents, preparing context, and steering AI to do the heavy lifting.The conversation covers practical ground for engineers at every level: how to successfully adopt agentic coding in your workflow, best practices for integrating tools like Codex into enterprise environments, and what it's really like to work at the frontier of AI deployment across industries like semiconductors, life sciences, and finance.Chapters00:00 Understanding the Role of Forward Deployed Engineers03:21 The Integration Process: Challenges and Solutions06:25 Optimizing AI Solutions with Codex09:38 Leveraging Codex for Team Efficiency12:28 Best Practices for Using Codex in Engineering Workflows15:29 Setting Up for Success in Enterprise AI Projects18:26 Navigating Stakeholder Engagement and Requirements21:16 The Future of AI in Enterprise Solutions25:53 Building Proof of Concept Solutions28:33 Collaborative Development and Model Improvement30:45 The Rise of Codex and User Adoption33:36 Integrating AI into Software Development36:10 Standardization vs. Customization in AI Tools39:05 The Evolving Role of Forward-Deployed Engineers42:48 Understanding the FDE Role at OpenAI46:10 The Recruitment Process at OpenAI49:50 Exploring Related Content49:58 Outro Final Coding Chats.mp4Arnaud's Linkshttps://www.linkedin.com/in/arnaudfrn/https://openai.com/index/introducing-openai-frontier/https://community.openai.com/t/introducing-the-new-codex-for-almost-everything/1379125https://openai.com/index/scaling-codex-to-enterprises-worldwide/John's Links:John's LinkedIn: https://www.linkedin.com/in/johncrickett/John’s YouTube: https://www.youtube.com/@johncrickettJohn's Twitter: https://x.com/johncrickettJohn's Bluesky: https://bsky.app/profile/johncrickett.bsky.socialCheck out John's software engineering related newsletters: Coding Challenges: https://codingchallenges.substack.com/ which shares real-world project ideas that you can use to level up your coding skills.Developing Skills: https://read.developingskills.fyi/ covering everything from system design to soft skills, helping them progress their career from junior to staff+ or for those that want onto a management track.

Coding Chats episode 76 - John talks to Laura Dietz - a computer science professor whose work focuses on whether AI evaluation metrics actually tell the truth. She's known for her critical take on "LLM as a judge" — not because she thinks it's useless, but because she wants numbers that mean something rather than numbers that just make a system look good.The conversation tackles some uncomfortable realities for software engineers: using an LLM to write code and another to review it is a circular trap, prompt engineering shouldn't be a computer scientist's day job, and every time you reject your code AI's output, you're quietly generating the training data that shapes its successor.Chapters00:00 Introduction to Laura Dietz and Her Journey03:12 Exploring LLMs as Judges06:16 Challenges in Evaluating Search Systems08:49 The Evolution of User Queries and Expectations11:46 The Role of LLMs in Information Retrieval14:44 Defining Quality in Search Results17:27 The Complexity of User Intent19:54 Human-AI Collaboration in Code Review22:53 The Future of LLMs in Software Development25:23 Balancing Human and AI Roles28:20 Innovative Approaches to AI Evaluation34:10 The Art of Assembling Ideas36:39 Balancing Cost and Quality in LLMs39:09 Evaluating LLM Performance43:50 The Future of LLMs and Training Data49:19 Exploring New Architectures in AI55:16 Understanding In-Context Learning01:00:45 The Role of AI in Creative Expression01:06:59 Exploring Related ContentLaura's Links:https://www.cs.unh.edu/~dietz/https://www.linkedin.com/in/laura-dietz-47036516/John's Links:John's LinkedIn: https://www.linkedin.com/in/johncrickett/John’s YouTube: https://www.youtube.com/@johncrickettJohn's Twitter: https://x.com/johncrickettJohn's Bluesky: https://bsky.app/profile/johncrickett.bsky.socialCheck out John's software engineering related newsletters: Coding Challenges: https://codingchallenges.substack.com/ which shares real-world project ideas that you can use to level up your coding skills.Developing Skills: https://read.developingskills.fyi/ covering everything from system design to soft skills, helping them progress their career from junior to staff+ or for those that want onto a management track.TakeawaysUsing an LLM to both generate and evaluate outputs is circular — like a student grading their own homework.If your evaluation metric can go up without your system actually improving, it's not a real metric.A better human-in-the-loop isn't one that rubber-stamps AI suggestions — it's one that's guided to look in the right place.LLMs don't get bored, which makes them genuinely useful for code review — but that's not the same as making them accurate."Faith-based engineering" — trusting AI output without validation — is a real and growing problem in software teams.Prompt engineering is a workaround, not a discipline; real engineers should be building systems, not crafting incantations.Every rejection you give your code AI is training signal — your frustration today is someone else's better tool tomorrow.The transformer attention mechanism is a weighted sum, and a sum isn't always the right operation — some problems need an AND, not an OR.AI tools are lowering the barrier to coding for people who were previously too intimidated to try, and that's worth celebrating.The same network effect that makes a platform valuable also makes monopoly in AI training data genuinely dangerous.