April 2025
April 30, 2025
Action Creates Gravity
April reinforced a principle that continues to serve me well: asking forgiveness rather than permission. Once you start moving, inertia works in your favor. People have endless opinions about why you shouldn't start something, but once momentum builds, they rarely try to stop the train. Speed and iteration beat pontification every time, though I definitely need more of that "train energy" myself – there's always more ground to cover.
The Perfect Digital Scratchpad
TLDRAW became my favorite discovery this month. Sometimes the simplest tools – especially ones with intuitive keyboard shortcuts – deliver the most power. Less friction equals more flow. The best tools disappear into your workflow rather than demanding attention.
Mental Muscle Memory
True learning feels less like downloading facts and more like building mental muscle. You need repetition to get basics on autopilot, freeing cognitive bandwidth for creative work. Applied this philosophy religiously to both product development and coding. Mastery isn't about knowing everything – it's about making fundamentals effortless so you can focus on the hard problems.
From Explore to Exploit
Shifted focus this month from exploration to exploitation by standardizing on a tech stack and building a reusable backend blueprint. Pure repetition builds speed and eliminates wheel-reinventing. Put the blueprint through three complete end-to-end projects – it's finally feeling like second nature. Templates and patterns are force multipliers for individual productivity.
The New Baseline Phenomenon
It's astounding how quickly cutting-edge becomes standard. Remember when massive LLM context windows felt like superpowers? Weeks later, they already feel... normal. As tools advance, so does the complexity of problems we throw at them. The goalposts keep moving, and we adapt at warp speed. This is the new normal: continuous recalibration of expectations.
End-to-End Product Realization
Built a complete product MVP (the UPF architecture) using AWS services (Sagemaker, S3) and internal data infrastructure (DataCentral, Redshift). 10,000+ lines of code, 30+ pages of documentation, and it's officially alive. Data flow, compute, storage – all integrated and functioning. There's profound satisfaction in bringing a complex system to life solo.
The Teaching Multiplier
Hosted a PM workshop at the Nashville offsite while simultaneously finalizing the UPF product build. Preparing 10+ hours of interactive training while completing complex technical work required serious juggling. Teaching forces you to understand your own work more deeply – the preparation time is an investment in your own understanding.
The Individual Revolution
Had a realization this month about cross-disciplinary capability. Work that previously required coordinating multiple specialists – software developer, product manager, data engineer, research scientist – is now soloable. Often faster too, with tight feedback loops and rapid iteration impossible when coordinating multiple people. This feels less like evolution and more like industrial revolution for knowledge work. The individual contributor just received a massive power-up.
Conquering Authentication
Authentication and Authorization – my personal Mount Everest of technical complexity – finally yielded to persistent effort. Previous attempts failed due to complexity and poor documentation. This time, modern LLMs served as technical sherpas, guiding through the landscape step by step. The difference wasn't more effort, but better guidance. Sometimes you need the right teacher, not just the right resources.
Simplicity Wins
Built a zip share forecast model that beats accuracy requirements and execution time criteria while maintaining full explainability. Funny how stepping back from complex neural networks and gradient boosting led to simpler approaches that performed better. Less is often more, particularly when transparency matters. Optimization isn't always about adding complexity – sometimes it's about finding the essential core.
Cursor as Universal Assistant
Started using Cursor extensively for non-coding tasks: drafting documents, file management, even diagram sketching. The real game-changer is its subtle "agentic" features – being able to command 'ls', 'mv', or 'touch' without breaking flow or switching to terminal. It cuts friction and context juggling dramatically. Less like talking to AI, more like having a hyper-efficient, context-aware assistant beside you.
April's central insight: individual capability is expanding exponentially through AI partnership. The constraint isn't technical knowledge anymore – it's imagination, initiative, and the willingness to act. When tools remove traditional bottlenecks, success flows to those who move fastest and learn most systematically. The future belongs to builder-learners who can orchestrate complex systems solo.