Ciklum

Another year, another Harvard Innovation Labs President's Innovation Challenge in the books! Congrats to the teams behind solar-powered oxygen delivery in Africa, AI for construction site risk, small molecules targeting chronic inflammatory disease, and many more. We spent the last few months going through the 2026 venture batch, and it felt a bit different from the last few cohorts: - 2023 was all about deep tech and biotech. Post-Covid, science and lab breakthroughs dominated - 2024, mostly GenAI wrappers. Some with more systems thinking and differentiation than others... - 2025: Agentic and edge AI deeply integrated into workflows. Robotic braiders, medical devices, diagnostics, all built as enhancers to existing systems. This year, a lot of teams are trying to rebuild whole verticals with AI-native domain-specific infrastructure. The platforms validate, monitor, coordinate, and plug straight into revenue and regulatory logic of their industries. We saw it in healthcare ops, compliance, therapeutics, a few others. What we noticed: - Unit economics and regulatory risk are coming up in pitches earlier than I'm used to - Fewer generic AI wrappers. More domain-native AI shaped by operational constraints of their industry - HealthTech teams in particular are baking reimbursement and compliance into product design from the start The finalists reflect this in different ways: Some are AI-native infrastructure, others are hardware-first or biotech-first. All of them treat the operational reality of their industry as the actual product. To everyone who made it into this year's PIC, and to the winners especially: Well done. #HIL #PIC2026 #Harvard #entrepreneurship #innovation #vc

Excited to kick off a new chapter! 🚀 I’m delighted to share that I’ve joined Blue Altair as a Consultant. Looking forward to collaborating with a talented team, taking on exciting challenges, and continuing to grow both personally and professionally. Here’s to new beginnings, continuous learning, and creating meaningful impact! #NewBeginnings #Consulting #CareerGrowth #ProfessionalJourney

Is your "Temporary Field" creating Permanent Debt? 🚩 We’ve all been there. You’re implementing a complex calculation, and you realize you need a place to store a value "just for a second." So, you add a nullable property to your Eloquent Model. It feels harmless. But it’s a classic Code Smell: The Temporary Field. The Problem A class should be a cohesive unit. If 3 out of 10 fields in your class are only used during one specific method call and remain null the rest of the time, you aren't writing OOP—you're creating a "Logic Junk Drawer." It makes the code harder to track, breaks encapsulation, and leaves other developers wondering: "Does this field have a value right now? Is it safe to use?" The Solution? Don't pollute the Model. Instead of making the data live on the object, pass it through a DTO or a dedicated Calculation Class. Why this matters for the long term: Predictability: Your Models stay as pure representations of your data. Testing: You can unit test the calculation without mocking a whole database object. Framework Agnostic: As I’ve discussed before, the more logic you pull out of the "magic" framework layers, the easier it is to maintain (and migrate!) in the future. Don't let your temporary fixes become permanent nightmares. Are you a fan of keeping Models "thin," or do you prefer having everything in one place for speed? Let's discuss in the comments! 👇 #PHP #Laravel #CleanCode #Refactoring #SoftwareArchitecture #WebDevelopment

After 2 editions mentoring tech students in the Electron hackathon hosted by the Faculty of Electronics, Telecommunications and Information Technology , this year I’m returning as a jury member for the AI & ML track and as the designer of the “Needle in Babel” challenge. 30 teams are building intelligent systems capable of reasoning, retrieving information, understanding, and accurately answering questions, while identifying traps across thousands of pages of noisy PDF documents. I strongly believe that knowledge retrieval is a key next-generation engineering skill in a future where information is more abundant than ever, with applications across all domains. I’m looking forward to discovering the innovative approaches students are building alongside their AI assistants. Thanks for the invitation, Teodora Mîrzan — great to see my former student driving such initiatives — as well as Alexandru ULMĂMEI and Bogdan George Carp , incredibly energetic tech enthusiasts and community event organizers in Romania. #InformationRetrieval #NaturalLanguageProcessing #ComputationalLinguistics #KnowledgeRepresentation #CognitiveSystems

Permits Shouldn't Slow You Down In the construction industry, the primary focus is on building, not navigating the complexities of permit applications. Builders often find themselves managing not only crews and client relationships but also the intricate process of securing permits. The challenges are significant: - Approval waits can stretch for weeks, leaving crews idle and increasing overhead costs. - Each municipality has its own set of requirements and forms, creating a constant learning curve. - Permit timelines are trending 10–20% longer in many regions. - More building departments are adopting complex digital submission systems that require precise formatting and documentation. Every correction cycle can cost valuable time and money that small builders simply cannot afford. It's time to streamline this process and focus on what truly matters: building.