Welcome to my blog!
This blog chronicles my journey of building things (currently interested in LLM-powered features) and figuring out life.
As a born designer who stumbled into data science before learning his first programming language (Python), I believe software is the (current) ultimate vehicle for great thinking and building.
But tools still need analytics to act as proof, and design to make them effortlessly digestible.
I live for these intersections, treat this as my playground for validating life thoughts by building them out, automating the (unfortunately essential) stuff to free up the fun.
Think of it as me trying to debug life while getting better at debugging programs.
Fueled by an insatiable curiosity (and AI's endless patience),
I'm pretty confident this entire script will be error-free by the end of this century.
Finished a refactor of the Dossigraphica research pipeline, switched the model from Gemini 3 Flash Lite to DeepSeek V4 Flash. Run first two tests on AMD and Amazon. The results are a fascinating window into where the pipeline stands today. To be honest, it’s not like the fix was intended to get the results that we are measuring now but after the first review of the tests, the completeness and robustness of data compared to previous version is impressive. This also led me to think a bit about measuring the output quality in an agentic extraction system.
If you’ve ever used Gemini’s deep research feature, you know the pattern: type a query, wait for the report, read through it, then manually extract the structured bits into JSON or markdown. I was doing this for corporate geopolitical analysis — office locations, supply chain dependencies, revenue geography, risk signals. Each analysis session meant copy-pasting, reformatting, and cross-referencing by hand. Tedious.
So I’ve been building something that does it for me.
Spent the last couple of nights (~20 hours) turning one of the “life outside building” ideas into an actual app.
It’s called Shotwise: upload a photo, get an AI critique, then generate an improved version.
This one was very different from my other project (Aperilex). Aperilex is a slow, deliberate learning grind; Shotwise is much more of a controlled dopamine sprint.
It’s been 7 weeks since I started building extensively with Claude Code. Now I have managed to deploy a working version of Aperilex which was a complete rewrite of the previous version which took about 5 months without deployment. Of course, there have been many ups and downs, but overall it’s been a fantastic learning journey. Time to review and reflect on the entire process.
Second week of Aperilex development, focusing on the application layer. Now that this layer is complete, it’s time to review the work done so far and reflect on the process. Fantastic and dream-like experience.
Been into a week of rebuilding Aperilex with Claude Code, focusing on the domain and infrastructure layers to build the solid foundation for the app. The experience has been intense, filled with learning and development challenges. Never thought about this kind of speed before but here we go. Today marks the completion for both phases and I’m excited to see how it all comes together.
Into the 4th day of rebuilding with Claude Code. Started to work on the edgartools and LLM in infrastructure layer overhauling the core of Aperilex. The main thrust was to pivot from generalized analysis to a highly structured, data-driven workflow, which involved a significant refactoring of our data ingestion and LLM orchestration layers.
Into the third night of coding with Claude Code, I focused on implementing the domain layer of Aperilex as outlined on the project PHASE.md, refining the core logic and data structures that will drive our SEC Filing Analysis Engine. This post details the process and decisions made during this phase.
As in the night 2 coding with Claude Code, I focused on establishing and refining the initial project setup for Aperilex while, of course, learning with the good practices provided by Claude (Opus4 and Sonnet4).
I knew this would come at some point as I was developing the SEC Analysis Project. Last weekend I started to refactor the codebase for better code quality check and planned the roadmap for better architecture and security. Instead of applying band-aid fixes, I decided today on a complete rewrite—but this time, of course, not alone.