Skip to content
podcast · research papersFellow

Research Insights Made Simple

A podcast about whitepapers, research and engineering methodology in plain language

Each episode reviews one study or important engineering topic: what is actually useful, where the methodology has limits, and how to apply the findings in real software development.

episode catalog

Episodes by topic

48 episodes · 1 recording pending
#482026-11-16

Developer productivity: how DX measures it

DX history, DX Core 4, AI Measurement and CAFE(S), with 2024–2026 PR throughput data and guidance on comparing the reports.

Developer productivityDevExAI in SDLCResearch methodology
2026-11-16recording pending

Agents over Data: Why SQL Is Not Enough

Агент над данными: почему одного SQL недостаточно

What a data agent needs beyond SQL: metric definitions, data lineage and access rights. The Snowflake and Databricks comparisons and a protocol for comparing harnesses on one model.

Data platformsAI in SDLCResearch methodologyPlatform engineering
#472026-11-10

AGENTS.md: Does a Context File Help the Agent?

Evaluating AGENTS.md: Are Repository-Level Context Files Helpful for Coding Agents?

How AGENTS.md affects coding agent behaviour, task success, and run costs on SWE-bench Lite and CTXbench.

AI in SDLCResearch methodology
#462026-11-09

How Much Harness Does a Strong Model Need?

Beyond the Model: Demystifying Harness Effects in Software Engineering Agents

When a strong model can use a simple harness and what changes on open-ended tasks, with evidence from NanoHarness and ML engineering agents.

AI in SDLCResearch methodology
#452026-11-09

Model or Harness: What Moves the Score?

Harbor Adapters and Harbor-Index

What changes when you swap an agent’s model or harness: evidence from 54 benchmarks, costs, timeouts, and the 82 Harbor-Index tasks.

AI in SDLCResearch methodology
#442026-11-02

Randomness in Agent Evals: Improvement or Noise?

On Randomness in Agentic Evals

Separating agent improvements from noise: repeated runs, zero temperature, pass@k and pass^k, trajectories, and evaluation budgets.

AI in SDLCResearch methodology
#432026-10-22

Agent Skills and Open Knowledge Format (OKF): Skills, Knowledge and Trust

Agent Skills and OKF with Maxim Smirnov: skill usefulness, evaluation, shared knowledge, RAG tradeoffs, and trust in agent outcomes. Guest: Maxim Smirnov.

AI in SDLCPlatform engineeringArchitecture governance
#422026-10-21

Agent evals: telling whether an agent succeeded

Demystifying evals for AI agents

How to evaluate AI agents: outcomes and trajectories, graders, pass@k and pass^k, isolated trials, and checking success criteria themselves.

AI in SDLCPlatform engineeringResearch methodologySRE / reliability
#412026-10-20

Agent Anatomy: Seven Harness Responsibilities

Harness Engineering: Anatomy, Architecture, and Evolution of Coding Agents. A Source-Code Study of Eleven Systems

Seven agent harness responsibilities: tools, context, permissions, isolation, and extensions. A review of eleven systems’ source code and the limits of its findings.

AI in SDLCPlatform engineeringArchitecture governanceResearch methodology
#402026-10-19

Anthropic: What Remains for People in a Company of Agents

How employment is changing at Anthropic

Anthropic's study of work with agents: hiring, productivity, human attention, engineering skills, and sharing the gains from automation.

AI in SDLCDeveloper productivityDevExResearch methodology
#392026-10-15

Scion: Anthropic’s Git Replacement for Agentic Development

Redesigning source control at Anthropic

Anthropic's Scion: server-side version control, caches, speedup limits, and questions about post-merge checks.

AI in SDLCPlatform engineeringSRE / reliabilityArchitecture governance
#382026-10-13

Harness Design for Coding Agents: What the Harness Contributes

An Empirical Study of Harness Design for Coding Agents

What planning, tools, and context management contribute to coding agents: 176 settings, cost, failure causes, and testing on your own tasks.

AI in SDLCDeveloper productivityPlatform engineeringResearch methodology
#372026-10-12

AI and the Data Platform: Connecting AI to All Organizational Data

Four approaches to data for AI agents: warehouses, federated queries, universal databases, and shared storage. Copies, lag, cost, and open formats. Guest: Nikolay Golov.

Data platformsPlatform engineeringDatabasesAI in SDLC
#362026-10-09

Harness Design for Long-running Application Development

Harness design for long-running application development

A planner, generator, and evaluator for long-running application development: acceptance criteria, browser checks, run costs, and simplifying the harness when models change.

AI in SDLCDeveloper productivityPlatform engineeringResearch methodology
#352026-10-08

Harness engineering: agent environments

Harness engineering: leveraging Codex in an agent-first world

How an OpenAI team built an environment for Codex: repository knowledge, observability, architecture checks, background cleanup, and the move to Symphony.

AI in SDLCDeveloper productivityPlatform engineeringResearch methodology
#342026-10-07

Long-running Agents: Work Handoffs

Effective harnesses for long-running agents

How an agent resumes work after a context reset: environment setup, a feature list, progress notes, Git, and browser verification.

AI in SDLCDeveloper productivityPlatform engineeringResearch methodology
#332026-10-05

SWE-agent: Interfaces Change Outcomes

SWE-agent: Agent-Computer Interfaces Enable Automated Software Engineering

A solo SWE-agent review: search, editing, validation and context, SWE-bench ablations, and the shift from a specialized interface to mini-SWE-agent.

AI in SDLCDeveloper productivityDevExResearch methodology
#322026-09-28

vLLM and PagedAttention: Virtual Memory for LLM Serving

Efficient Memory Management for Large Language Model Serving with PagedAttention

A review of the vLLM and PagedAttention paper: why LLM serving is bound by KV cache memory, how paged virtual memory from operating systems removed reservation and fragmentation, and how that bought 2–4× throughput without changing the model.

Platform engineeringArchitecture governance
#312026-09-22

AI4SDLC: What I Would Do Differently

What to rent and what to own in an AI stack: agent loops, context, permissions, tools, and verifiable engineering tasks. The extended Deep Tech Night talk.

AI in SDLCPlatform engineeringArchitecture governanceDeveloper productivity
#302026-09-21

Developer Productivity: Google’s Human-Centered View

Developer Productivity for Humans

Developer Productivity for Humans: goals, measurement, builds, onboarding, quality, teamwork, and AI tensions through Google research and Alexander Polomodov’s reviews.

Developer productivityDevExResearch methodologyTechnical debtAI in SDLC
#292026-08-27

AI-Native SDLC: Code Accelerated, Delivery Didn't

The AI-Native SDLC Playbook

A review of the AI-Native SDLC Playbook: redesigning planning, design, build, test, deployment, and maintenance around AI agents, versioned artifacts, and human control points. Guest: Anton Kosterin.

AI in SDLCDeveloper productivityArchitecture governanceSecurity
#282026-08-24

The Data Platform in 2026: From DWH to Lakehouse and AI Agents

How to separate OLTP from OLAP, where MPP warehouses reach their limits, what makes a lakehouse, why legacy migration takes years, and how AI agents change data-platform requirements. Guests: Nikolay Golov, Alexander Filatov.

Data platformsPlatform engineeringDatabasesAI in SDLC
#272026-08-07

AI Development as a Co-Evolving Stack

AI Development as a Co-Evolving Stack: Hardware, Models, Harnesses, Tools, and Traces

How hardware, models, harnesses, tools, traces, and evals form a co-evolving AI-development stack, and where enterprises should draw the rent, adapt, and own boundary.

AI in SDLCPlatform engineeringArchitecture governanceResearch methodology
#262026-07-31

Building Evals That Work

Production-grade evals for AI agents

How to turn AI-agent evals from one-off answer grading into a reproducible engineering system with replayable episodes, hidden judges, traces, scorecards, and release gates. Guest: Evgeny Sergeev.

AI in SDLCDeveloper productivityResearch methodology
#252026-07-30

Modeling Reliability from a Dependency Graph

Model Discovery and Graph Simulation: A Lightweight Gateway to Chaos Engineering

How to derive a dependency graph from traces, estimate availability with Monte Carlo, and prioritize chaos experiments. Guest: Anatoly Krasnovsky.

SRE / reliabilityPlatform engineeringResearch methodology
#242026-07-29

Why the Architect AI Copilot Still Hasn’t Arrived

Artificial Intelligence Support for Software Architecture Practice

What AI can do in software architecture, where context breaks, and why humans remain accountable for architectural trade-offs. Guest: Sergey Baranov.

AI in SDLCArchitecture governanceResearch methodology
#232026-07-23

The Economics of AI Development: From Tokens to Accepted Work

Why cheaper tokens do not guarantee a smaller AI budget, and how to manage cost per accepted task, trace budgets, routing, and local-model TCO.

AI in SDLCDeveloper productivityPlatform engineeringArchitecture governance
#222026-07-20

How to Build a Governed Agent Stack

Конфигурации агентного стека: обвязка × модель × инструменты

How to choose a harness, model, and tools, define data and authority boundaries, and turn an agent stack into a governed enterprise system. Guest: Mikhail Trifonov.

AI in SDLCPlatform engineeringArchitecture governanceSecurity
#212026-07-17

Measuring Coding Agents in Dialogue

SWE-Together × SWE-INTERACT

A review of SWE-Together and SWE-INTERACT with Aleksey Litvinov: measuring coding agents when requirements emerge during the work. Guest: Aleksey Litvinov.

AI in SDLCDeveloper productivityResearch methodology
#202026-07-16

Why Coding Agents Do Not Eliminate Expertise

Agentic Coding and Persistent Returns to Expertise

A review of Anthropic's research with Evgeny Sergeev: why coding agents amplify domain expertise rather than eliminate it. Guest: Evgeny Sergeev.

AI in SDLCDeveloper productivityResearch methodology
#192026-07-14

Loop Engineering: Designing Systems That Run Coding Agents

Loop Engineering: The Anthropic Playbook for Designing Systems That Prompt Your Agents

A review of Loop Engineering with Maxim Smirnov: agent loops, generator/evaluator, hidden debts, and safe autonomous-system design. Guest: Maxim Smirnov.

AI in SDLCDeveloper productivityArchitecture governance
Available on the site
49
episodes
1
recording pending
2024-2026
episode archive
201
Yandex Music subscribers

Why listen

  • 01Understand whitepapers faster before deciding whether to read the full text
  • 02Separate strong research findings from methodological constraints
  • 03Connect research findings with engineering practice: AI in SDLC, DevEx, productivity and governance
  • 04Build a map of ideas for architecture and organizational decisions in large-scale software development

Episode format

One episode, one paper

Each episode is anchored in a specific research paper, report or whitepaper.

Methodology matters

We look at study design, samples, limits, metrics and what should not be overgeneralized.

Practical translation

Academic and industry research gets translated into engineering decisions, process changes and management trade-offs.

Topics covered

The episodes follow several lines: productivity, AI4SDLC, governance and platform thinking.

01Developer productivity
02DevEx and flow
03AI in SDLC
04AI code assistants and agents
05DORA / SPACE / DX Core 4
06API governance
07Architecture governance
08Technical debt
09Security by design
10Data platforms
11Database architecture
12Research methodology

Where to listen

The podcast is available in audio and video - pick the platform that fits.

Who it's for

  • Engineers and architects who want research without academic noise
  • Engineering managers and CTOs making decisions about productivity, DevEx and AI tooling
  • Platform teams looking for arguments around governance and internal standards
  • People who want to read papers more deliberately and find applicable ideas faster

Listen to the first episode

49 whitepaper and engineering research reviews. Start with API Governance at Scale or pick a topic in the episode catalog.

Open the YouTube playlist