Sandro Volpicella

𝗔𝗿𝗲 𝘁𝗲𝗰𝗵𝗻𝗶𝗰𝗮𝗹 𝗵𝗶𝗿𝗶𝗻𝗴 𝘁𝗲𝘀𝘁𝘀 𝘀𝘁𝗶𝗹𝗹 𝘂𝘀𝗲𝗳𝘂𝗹, 𝘄𝗵𝗲𝗻 𝗰𝗵𝗲𝗮𝘁𝗶𝗻𝗴 𝗵𝗮𝘀 𝗯𝗲𝗰𝗼𝗺𝗲 𝘁𝗵𝗲 𝗻𝗼𝗿𝗺? Last week, we had a great conversation - thanks again to Harald Huber from RETENCON AG and Maria Lohwasser, Patrick Schianchi, Christian Greb - about the growing problem of cheating in software development hiring tests. Let’s be honest: With tools like Claude, Gemini, Cursor, it has become incredibly easy to cheat on remote coding challenges. And many candidates do. That raises some uncomfortable but necessary questions: - How do we fairly and accurately assess real skills today? - How do we detect not just technical ability, but collaboration, problem-solving, and team fit? From my perspective: ✅ Trial projects or structured test days with real team members are much more insightful yes, it takes time, but it’s worth it. ✅ Time-pressured “multiple choice tests” can reduce the chance of using AI but they only scratch the surface. ✅ Ultimately, we should shift our focus: from can this person write code to can they think in systems and work with others? Because let’s face it: If machines are writing the code - what exactly do we expect from junior developers? Software hiring needs to evolve. Fast. Not more algorithm quizzes. More real-world thinking, communication, and architectural sense. I’d love to hear how you’re adapting your hiring practices to deal with AI-assisted cheating. Is this something you’re already facing?

24

11 Kommentare

𝗪𝗵𝗮𝘁 𝗜’𝗱 𝘀𝘂𝗴𝗴𝗲𝘀𝘁 𝗶𝗳 𝘆𝗼𝘂’𝗿𝗲 𝗮 𝗖𝗧𝗢 𝗼𝗿 𝗧𝗲𝗰𝗵 𝗠𝗮𝗻𝗮𝗴𝗲𝗿 𝘀𝘁𝗿𝘂𝗴𝗴𝗹𝗶𝗻𝗴 𝘁𝗼 𝗵𝗶𝗿𝗲 𝗹𝗼𝗰𝗮𝗹𝗹𝘆: -Don’t lower the bar just to make a hire -Look at where the real tech talent hubs are -Plan collaboration properly (same city, co-working, hybrid setups) I’ve just done this with a German AI commerce client. They needed 3–5 backend engineers, but the local market was expensive and short on the right skills. We built the team in Spain instead. Within 6 weeks, 4 senior contractors onboarded, all set up in the same city so collaboration stays tight. 👉 The best engineers aren’t always neccasarily where you are.

6

Empfehlenswertes Paper: "Measuring Agents in Production" https://lnkd.in/dKFpvXzq Bestätigt alle meine eigenen Beobachtungen. "Echte" Agents-Systeme sind schwer und KISS auch bei AI ein guter Ansatz. Was mich überrascht hat: "85% (Interviews) nutzen "Custom Code" statt Frameworks (wie LangChain/CrewAI), um die Kontrolle zu behalten." Hier die Findings vom Paper:  RQ1: Why build agents? Productivity drives adoption. Organizations deploy agents primarily to automate routine tasks and reduce human task hours, prioritizing measurable efficiency gains over novel capabilities. • RQ2: How are agents built? Simplicity and controllability dominate. Production systems favor closed-source models utilizing manual prompting rather than weight tuning. Architecturally, these systems rely on structured workflows with bounded autonomy and typically execute limited steps before human intervention. • RQ3: How are agents evaluated? Human verification remains the primary method. Practitioners rely heavily on human-in-the-loop evaluation because clean baselines and ground truth datasets are scarce. Automated techniques like LLM-as-a-judge complement human review rather than replace it. • RQ4: What are the challenges? Reliability represents the central development focus. The difficulty of ensuring correctness and evaluating non-deterministic agent outputs drives this friction. Latency and security typically act as manageable constraints rather than hard blockers as engineering workarounds and restricted environments currently manage them. #stillearly #decadeofagents #ai #agents

15

Last Thursday, I held my first session at the "AI Software Engineering Bootcamp" 🤖 of Digitale Leute School. With AI, the capabilities and responsibilities of engineers are widening. Engineers can - and need to be - closer to the customers and to the product. That is why the first module is about Product Management. 🚀 In my session about "Stakeholder Management and Storytelling", we dove deep into: → Reading the room: How to address stakeholders in different kinds of companies (business-led vs. tech-led vs. product-led) → Prioritizing with spine: Prioritization frameworks that work → Working backwards: Using PR/FAQs to align stakeholders around big bets by turning complex products into simple stories → Saying "no" without burning bridges: How to do your most important job without risking relationships Despite being three hours in the evening, I was very happy about the energy and engagement of the participants. 🔋 My next session is coming up in Feb and will be about "Coding with AI responsibly". Looking forward!

62

8 Kommentare

The AI marketplace moves fast. New tools land every week—each one promising disruption, productivity gains, and competitive edge. But speed isn’t the same as strategy. Installing the latest plugin doesn’t equal innovation. In fact, mistaking novelty for necessity is how organizations end up scaling noise instead of value. The real innovators don’t ask “what’s new?” They ask “what matters most right now?” They move when the technology aligns with their vision, when it strengthens their unique edge, and when it solves problems that actually matter. Hype fades. Alignment endures. Read more about it in my latest article:  🇺🇸 EN: https://lnkd.in/grNGn4Gy  🇩🇪 DE: https://lnkd.in/gu6mAREh #AIImpact #AIInsights #AIEverywhere #AIStrategy #AIDoneRight #DigitalTransformation #InnovationLeadership

9

🎙️ Podcast weekend edition: Moving from "AI as a helper" to "AI as a doer" – Exploring Agentic Coding with Straion 💻🚀 Is the way we build software being rewritten in real-time? In the latest episode of AI Talk, Jakob and I are thrilled to welcome Katrin Freihofner, founder of the Linz-based startup Straion. Having recently closed a funding round, Katrin and her team are diving deep into one of the most dynamic fields in tech: Agentic Coding. In this episode, we discussed: 👩‍💻 Beyond Copilot: Why Straion focuses on the "agentic" part of development and how it differs from the tools we’ve seen so far. ⛰ The DACH Startup Edge: How a lean, focused team from Austria can compete in a global market dominated by Silicon Valley giants. 🔦 The "Flashlight App" Problem: Will specialized software tools survive, or will general-purpose agents eventually absorb them all? 🤝 Future-Proofing Talent: Why human empathy, strategic thinking, and "common ground" remain the ultimate primitives of success, even as the code writes itself. I’m already looking forward to our follow-up talk for their Series B next year. /jk Their speed of execution in this space is breathtaking. Watch the full interview and read our deep-dive article here: 📺 https://lnkd.in/d3vs2_J4 Lukas Holzer Matthias Seiderer Andreas Greilhuber Fabian Friedl Trending Topics by newsrooms newsrooms Adam Ehrlich Matthias

42

1 Kommentar

𝐓𝐡𝐞 𝐆𝐞𝐫𝐦𝐚𝐧 𝐓𝐞𝐜𝐡 𝐓𝐚𝐥𝐞𝐧𝐭 𝐂𝐫𝐢𝐬𝐢𝐬: 𝐐𝟒 𝟐𝟎𝟐𝟒 𝐃𝐚𝐭𝐚 𝐀𝐧𝐚𝐥𝐲𝐬𝐢𝐬 After analyzing 847 SME and startup hiring processes across Bavaria, Berlin, and NRW in Q4 2024, the bottleneck is worse than reported. 𝐓𝐡𝐞 𝐑𝐞𝐚𝐥 𝐍𝐮𝐦𝐛𝐞𝐫𝐬: 149,000 unfilled IT positions (Bitkom, Sept 2024) But drill down by role: • Senior Backend: 11.2 months average vacancy • DevOps Engineers: 9.8 months • Frontend React: 6.4 months 90% talent shortage statistic? It's actually 94% for companies with 50-250 employees in tech hubs. 𝐓𝐡𝐞 𝐇𝐢𝐝𝐝𝐞𝐧 𝐂𝐨𝐬𝐭 𝐒𝐭𝐫𝐮𝐜𝐭𝐮𝐫𝐞 Internal hiring cost breakdown (50-person startup, Berlin): • HR/recruiting FTE: €67K + benefits • Average 47 interview hours per senior hire • Opportunity cost: €23K per unfilled month • Contractor premium: 180-220% of FTE cost Frankfurt automotive supplier (127 employees): "We spent €340K on contractors in Q3 because we couldn't fill 2 permanent roles." 𝐑𝐞𝐠𝐢𝐨𝐧𝐚𝐥 𝐕𝐚𝐫𝐢𝐚𝐭𝐢𝐨𝐧𝐬 𝐌𝐚𝐭𝐭𝐞𝐫 Berlin: 8.2 months avg time-to-hire, salary expectations +15% YoY Munich: 9.7 months avg, +23% salary inflation (automotive demand) Hamburg: 6.8 months avg, better conversion rates Stuttgart: 11.1 months avg, Mercedes/Porsche competition Data from 340 placements tracked Oct 2023 - Oct 2024 𝐖𝐡𝐚𝐭'𝐬 𝐀𝐜𝐭𝐮𝐚𝐥𝐥𝐲 𝐖𝐨𝐫𝐤𝐢𝐧𝐠 Analyzed successful scaling cases from our client portfolio (10-50 dev teams): Pattern 1: Dedicated Team Model (89% success rate) • Pre-vetted developers integrated as extended team • Full German project management oversight • Cultural alignment workshops in month 1 • Result: 3.2 months scaling vs 14.8 months local hiring Pattern 2: Hybrid Integration Approach (78% success rate) • 2-3 local leads for architecture/strategy • Remote dedicated developers for execution • Single point of contact for management • Cost reduction: 50-60% vs German FTEs Pattern 3: Rapid Team Scaling (emerging) • From 5 to 25 developers in 8 weeks • Pre-existing talent pools, no recruitment delays • Retention: 91% stay 24+ months vs 64% local market • Growing trend among scaling startups in our network 𝐓𝐚𝐜𝐭𝐢𝐜𝐚𝐥 𝐑𝐞𝐜𝐨𝐦𝐦𝐞𝐧𝐝𝐚𝐭𝐢𝐨𝐧 If you're scaling tech teams in Germany (or Western Europe): Stop optimizing hiring process. Start optimizing talent access. The companies solving this aren't hiring faster locally. They're building distributed teams strategically. Based on 2+ years helping German SMEs and startups scale, the pattern is clear: traditional hiring is broken, but the alternative isn't just "remote workers." It's dedicated team partnerships with proper integration. What's your biggest scaling bottleneck right now?

22

6 Kommentare

About a week ago, I attended an AI coding event in Hamburg. People’s dopamine levels went through the roof because they could finally build what they loved. But the problems started when it became clear that most participants didn’t know even the basics of programming. AI coding shouldn’t be just a dopamine booster. I’m trying to explain the fundamentals of programming concepts in plain, human language because if you want to build serious products (not just funny little toys), you must understand the basics. read "Cookie" full article here: https://lnkd.in/e57e7mHW

35

4 Kommentare

🚨🚨 WE ARE PIVOTING 🚨🚨 Denzell and I are moving on from search. Unlike when we started in 2023, there are now many different open source solutions for quickly building highly relevant retrieval which I wouldn't mind recommending. Trieve is a great one, we have hundreds of paying customers, serve millions of requests every day, and make hundreds of thousands of dollars. But it's done. We have completed the project. Listed on AWS marketplace, built a drop-in component, shipped a Shopify app. There's no product left to build. Sure, we could grind more SEO and directory listings until we hit millions in revenue. But we'd burn out long before that happens, and frankly, that's not where we win. Our edge isn't in outmarketing Elastic or out-enterprise-selling OpenSearch. It's in being deeply technical and shipping fast. Different game entirely. If you use Trieve and like it, don't worry, it's not going anywhere. We'll still be resopnding to bug reports and feature requests. Same SLA, same support quality. It keeps us profitable! Just don't expect any big new features or flashy launches. Our new thing is called Threshold Security (onthreshold . com). Open source alternative to Coinbase Custody. Consider this, BITB is a $4B BTC ETF that stores everything at Coinbase. Forget crypto, I wouldn't put $4B of fiat USD into a single bank. I get why Bitwise does it though. Setting up your own custody solution is a nightmare rought now. Libraries exist (shouts out zcash and Luke Parker), but nobody has made a good server implementation. Coming from the search world, it's like having Lucene with no ElasticSearch. Soooo, we're fixing that! Going to build the thing which should already exist.

163

19 Kommentare