Kanalist
DEVCLUB.EE keskendub tarkvaratöö otsustele, mis ulatuvad üksiku koodirea kirjutamisest kaugemale. Kuupäeva, esineja nime ning teemaga tähistatud ettekanded käsitlevad süsteemide toimimist, testimist…
DEVCLUB.EE keskendub tarkvaratöö otsustele, mis ulatuvad üksiku koodirea kirjutamisest kaugemale. Kuupäeva, esineja nime ning teemaga tähistatud ettekanded käsitlevad süsteemide toimimist, testimist, andmebaase, disaini ja meeskonna koostööd. Valikus kohtuvad tehnilised lahendused ning töökorralduslikud küsimused, mistõttu saab arendamist vaadelda tervikuna: kuidas süsteem valmib, kuidas seda kasutatakse ning kes selle toimimise eest vastutab.
Näiteks ebastabiilseid ühikteste käsitlev ettekanne suunab tähelepanu testitulemuste usaldatavusele. Valvekorralduse teema avab arendaja vastutust pärast tarkvara kasutuselevõttu, pangas skaleeruvate mikroteenuste käsitlus toob sisse konkreetse tegevuskeskkonna. Need näited aitavad liikuda tehnoloogia valimisest selle igapäevaste tagajärgedeni. Küsimus pole ainult selles, kas lahendus töötab, vaid ka selles, kuidas selle tööd kontrollida ning probleeme lahendada.
Tehisintellekti agentide selgitus ja MCP tööriistade kasutamine andmebaasitöös lisavad valikusse uusi tööviise. Kasutajaliidese disainitaju, teenusedisaini ning avalike suhtluskanalite eelistamise teemad laiendavad vaadet veelgi. Ettekannete kaudu saab jälgida, kuidas tehniline kvaliteet seostub kasutuskogemuse või sellega, kuidas teadmine meeskonnas liigub.
Kanal sobib arendajale, testijale, tehnilisele juhile või disainerile, kes soovib mõista ka enda põhirollist väljapoole jäävaid otsuseid. Mõni osa pakub võimalust süveneda tuttavasse probleemi, teine aitab sõnastada küsimusi koostööpartnerile. Ettekandeformaat võimaldab keskenduda ühele teemale korraga ning võrrelda eri esinejate rõhuasetusi. Läbiv väärtus on võimalus siduda tehnilised valikud süsteemi kasutamise, hooldamise ja meeskonna tööga.
Uusim osa
2026.10.29 Sreekanth Harigovindan - The EU AI Act: Compliance Is an Engineering Problem
EU AI Act compliance is an engineering problem, not just a legal one. If you ship features on GPT, Claude, or Gemini, your logs, code reviews, and docs are your compliance evidence.
SDET Sreekanth…
EU AI Act compliance is an engineering problem, not just a legal one. If you ship features on GPT, Claude, or Gemini, your logs, code reviews, and docs are your compliance evidence.
SDET Sreekanth Harigovindan (Riverty) uses his open-source AI test generator as a live case study: classifying it under the EU AI Act risk tiers, mapping repo artifacts to Articles 11–14, 50, and 72–73, and explaining what changes for engineering teams between December 2026 and 2028.
🔑 TL;DR
• Classify before you build: inventory every AI system (SDKs, model versions, vendors), document its purpose, assign a risk tier (prohibited / high-risk / transparency / minimal), and define your role: provider, deployer, or both.
• Much of the evidence already lives in your SDLC: CI tests, code review, version history, and monitoring map to record-keeping (Art. 12), human oversight (Art. 14), and technical documentation (Art. 11/13).
• The gaps are AI-specific: approval and rejection records with trace IDs, prompts and SKILL.md files versioned in Git, AI-content labelling (Art. 50), and an incident plan for agents running unattended (Art. 73).
• Case study: a human-in-the-loop approval gate, HTML replay of agent actions, and LLM-as-judge scoring (0–5), plus honest trade-offs like a shared vector store and a root Docker image.
📅 EU AI Act milestones covered
• Feb 2025: prohibited practices + AI literacy
• Aug 2025: general-purpose AI (GPAI) model obligations
• Aug 2026: general date of application
• Dec 2026: marking/labelling of AI-generated content
• Dec 2027: stand-alone high-risk AI (hiring, credit scoring, education)
• Aug 2028: high-risk AI embedded in products (machinery, medical devices)
Per Regulation (EU) 2024/1689 as amended by the Digital Omnibus on AI, Regulation (EU) 2026/1744.
⏱️ Chapters
0:00 Compliance as an engineering problem
1:34 Warm-up: why don't prompts get regression tests?
3:07 EU AI Act timeline: 2025 → 2028
4:02 "Made with AI": content labelling from Dec 2026
5:16 The four risk tiers
6:01 Scope: does the Act apply outside the EU?
7:16 Risk and human-in-the-loop
7:45 Roles: provider vs. deployer (and being both)
8:32 Mapping engineering practices to compliance evidence
9:31 Approval records, trace IDs & log retention
10:29 Incident plan for autonomous agents
11:36 Building an AI inventory: SDKs, models, vendors
12:17 7-step starter compliance checklist
13:44 Reading the regulation: 2024/1689 & 2026/1744
14:55 Mapping repo docs to Articles 11–14, 50, 72–73
15:57 Machine unlearning & what to store in a vector DB
18:04 Case study: classifying an AI test generator
19:29 Human approval gate, HTML replay & LLM-as-judge
20:49 Workflow walkthrough & known trade-offs
22:01 Before you sell AI in the EU: checklist
23:36 Q&A: Prompt & SKILL.md traceability in Git
26:18 Q&A: How to label AI-generated content
27:11 Q&A: What to log: prompts, outputs, reasoning?
29:44 Q&A: Fine-tuning and change records
30:50 Q&A: Fines and penalty tiers
31:44 Q&A: Prompt versioning with LangSmith & LangGraph
32:30 Q&A: When to report a serious incident
34:36 Practitioner view: building the AI inventory with legal
35:31 Q&A: Is compliance ever "done"?
37:48 Q&A: Why build this open-source project?
40:30 Q&A: Is AI-assisted hiring always high-risk?
🛠️ Tech stack & references
• Regulation: EU AI Act (Regulation (EU) 2024/1689), Digital Omnibus on AI (Regulation (EU) 2026/1744)
• LLM providers: OpenAI GPT, Anthropic Claude, Google Gemini
• Orchestration & prompt versioning: LangChain, LangGraph, LangSmith
• Pattern memory: vector store (FAISS + SQLite)
• Test frameworks: Playwright, Cypress, WebdriverIO, Appium
• Distribution: Docker (GHCR), Hugging Face Spaces
• Observability: trace IDs, OpenTelemetry, Grafana Loki
• License: AGPL-3.0
🔗 Links
Project: https://github.com/aiqualitylab/ai-natural-language-tests
Live demo: https://demo.aiqualitylab.org/
Speaker: https://www.linkedin.com/in/sreekanthharigovindan/
EU AI Act (official text): https://eur-lex.europa.eu/eli/reg/2024/1689/oj
Digital Omnibus on AI: https://eur-lex.europa.eu/eli/reg/2026/1744/oj
DevClub: https://devclub.ee
👤 About the speaker
Sreekanth Harigovindan is an SDET/QA engineer at Riverty. He writes about AI in quality engineering and maintains ai-natural-language-tests, an open-source platform that turns natural-language requirements into Cypress, Playwright, WebdriverIO, and Appium tests.
⚠️ This talk is an engineering perspective, not legal advice. Article numbers and deadlines can change, so confirm obligations with your legal/compliance team.
#EUAIAct #AICompliance #SoftwareEngineering
Subscribe: https://www.youtube.com/channel/UC9lRNfzenw_7MVmDPFyRFjQ?sub_confirmation=1
Our website: https://devclub.ee
Our blog: https://blog.devclub.ee
Facebook-group: https://www.facebook.com/groups/devclub.ee
YouTube Shorts
DEVCLUB.EE - EESTI KEELES
DEVCLUB.EE - IN ENGLISH
DEVCLUB.EE - live streams
DEVCLUB.EE 2011 (22 of 24)
DEVCLUB.EE 2012 (31 of 33)
DEVCLUB.EE 2013 (14 of 18)
DEVCLUB.EE 2014 (16 of 17)
DEVCLUB.EE 2015 (12 of 18)
DEVCLUB.EE 2016 (9 of 9)
DEVCLUB.EE 2017 (15 of 18)
DEVCLUB.EE 2018 (25 of 25)
Uploads
Rohkem selle kanali osi
2026.10.29 Andrei Skopenko - AI Second Brain for Modern Project Management
8.11.2026
AI second brain for project management: how do you preserve the decisions, ownership, and context scattered across tickets, chats, and documents? Andrei Skopenko, VP of Engineering at Wallester, shares his WIKI-LLM approach to building...
2026.09.17 Franklin Ezeji - SchedNext: Building a Lightweight Agent Runtime for Cloud, Edge, and...
5.10.2026
SchedNext is a lightweight agent runtime for cloud, edge, and embedded Linux. Learn how it schedules, executes, and exposes workload state without a database, always-on web API, or heavy orchestration stack. An "agent" here...
2026.09.17 Zeynep Öztuğ - Thinking Beyond Code: Practical Engineering Decisions in Modern Systems
26.9.2026
Practical software engineering decisions require more than code: they require diagnosing the right problem, weighing trade-offs, and owning the outcome. Zeynep Öztuğ, Engineering Lead at Bally's Interactive, uses a slow API and a production...
2026.06.30 Elis Laasik - Train your design eye (UI design eye)
29.8.2026
UI design for developers becomes easier when you know what to inspect. Elis Laasik shows how to train your design eye and diagnose interfaces that feel “off” using four repeatable rules: alignment, spacing, sizing,...
2026.06.30 Umut Pazarbasi - So you want to be a DBA?
18.7.2026
Thinking about what it takes to excel as a Database Administrator (DBA), or looking to bridge the technical gap between application development and database reliability? This technical session deep-dives into the real-world operational challenges,...
2026.04.22 Dennis Garavsky - Why I recommend Public Channels over Private Chats
10.5.2026
Stop losing 80% of project context to private silos. Dennis Garavsky (Principal PM/TPM at Developer Express, Inc) explains why migrating organizational communication to public channels is a technical necessity for high-performance engineering teams and...
2026.04.22 Vladimir Ivanov - What are AI agents?
2.5.2026
What are AI agents? This video gives a concise, engineering-focused breakdown of agentic AI: the software patterns, runtime components, and engineering tradeoffs that make an LLM-driven system behave like an autonomous agent. Vladimir Ivanov,...
2026.03.04 Andrei Solntsev - Flaky unit-tests
21.3.2026
While the industry often blames UI tests for instability, flaky unit tests represent a more insidious threat to CI/CD pipelines and developer trust. In this session, Andrei Solntsev challenges the assumption that mocking external...
2026.03.04 Andrei Skopenko - You Build It, You Run It: The Pain & Power of On-Call
7.3.2026
Is the "You Build It, You Run It" model the secret to high-performing teams or just a recipe for burnout? In this session, Andrei Skopenko (VP of Engineering at Wallester) deep dives into the...
2026.02.04 Oluwafemi Shobande - Durable Execution
1.3.2026
Stop losing state when your services crash. 🛠️ In this session, Oluwafemi Shobande (Senior Software Engineer at Bolt) breaks down the power of Durable Execution—a paradigm shift for building long-running, fault-tolerant workflows that survive...
2026.02.04 Andrei Gornostajev - Software Development in a Regulated Environment
21.2.2026
Discover how to align secure software development with heavy financial regulations without slowing down your engineering team. Andrei Gornostajev, CTO at Wallester, breaks down practical FinTech compliance for IT professionals. In this DevClubEU talk,...
2025.12.18 Markus Backman - Andromeda: Opinionated Micro-services That Scale at a Bank — our journey
18.1.2026
How do you scale microservices in a highly regulated banking environment without falling into operational chaos? In this session, Markus Backman (Technology Enterprise Domain Architect at Swedbank) reveals the journey of building Andromeda, an...
2025.11.18 Anna Shishkova - Service Design in Action: From Concepts to Real-World Impact
30.11.2025
Service Design in action – discover how to turn service concepts into real-world impact for your organization. In this talk, Anna Shishkova (Product Business Analyst) explains what Service Design is, who it’s for, and...
2025.10.07 Bobur Umurzokov - MCP tools as your database engineer
9.11.2025
Stop losing 20% of your development time to manual database tasks. In this DevClubEE talk, Bobur Umurzokov demonstrates how to use MCP tools and AI agents as your personal "AI Database Engineer" to automate...
2025.10.07 Vladimir Ivanov - Building solutions that work
1.11.2025
Stop building products that nobody uses. In this talk, tech leader Vladimir Ivanov shares the critical product discovery principles that most development teams ignore. You'll learn why so many projects fail by rushing into...
2025.09.03 Olga Khodyreva - Teaching IT in schools
20.9.2025
Is teaching IT in schools just about boring theory? Think again. Olga Khodyreva takes us inside the modern Estonian classroom to reveal the methods that actually get children passionate about programming, design, and creating...
2025.09.03 Dmytro Kolomiets, Külli Siimon - Engineer 2.0: The Playbook Big Tech Wrote – Now You...
13.9.2025
Is your engineering skillset up-to-date? Big Tech companies like Amazon, Google, and Microsoft have quietly changed the rules for what makes a successful engineer, and these changes are now affecting everyone. This is the...
2025.06.30 Oleg Golovachov - Why Managers Aren't As Dumb As We Thought
20.7.2025
Ever felt like managers just don't get it? Oleg Golovachov, a seasoned tech leader who journeyed from developer to CTO, offers a candid insider's perspective in this DevClubEE talk. Discover why management decisions often...
2025.06.30 Andrei Solntsev - Why do we have 7 notes?
5.7.2025
Unravel the fascinating mystery behind Western music's 7-note scale! Why seven, and not six or eight? Join Andrei Solntsev, a renowned software developer and author of Selenide, as he demystifies this fundamental question using...
2025.03.26 Stanislav Belyaev - Authority, Power, and Expertise: How Do They Shape Your Work?
12.4.2025
Are you curious how authority can redefine your career and boost your professional success? In this eye-opening video, we break down the true impact of authority on power, influence, and decision-making across today's dynamic...
Innohub TV
Vaata seda sisu ka Innohub TV-s
Valisime sulle Innohub TV-st klipid, mis jätkavad sama teemat.
