AI Retrospective — A Year of Building, Systems, and Compounding
A recruiter-facing retrospective on how I used AI to think, build, and ship real products.
This is not a highlight reel. It is a technical and strategic retrospective on how I used AI as leverage to build real things.
The Identity: Builder • Technologist • AI-First Thinker
In 2025, I shifted from asking "How do I learn this?" to "How do I build this properly?" This retrospective exists to help recruiters and builders understand how I operate when given leverage.
Primary Roles Played:
- Product Architect: Designing systems for real-world applications.
- AI Prompt Engineer: Mastering the art of instruction to remove friction.
- Full-Stack Consultant: Production-grade engineering and technical strategy.
My AI Operating Model
I do not use AI to replace thinking. I use AI to remove friction between intent and execution.
[!IMPORTANT] AI is most powerful when paired with clarity. I treat AI as a clarity engine, not an answer engine.
The Default Loop:
- Define the Problem: Start with the real-world problem, not the feature request.
- Structure Over Answers: Use AI to decompose vague goals into constraints and identified assumptions.
- Immediate Execution: Build the prototype before the theory becomes heavy.
- Encoding Systems: Use AI to identify second-order effects and encode learnings into repeatable systems.
Proof of Work: Where AI Created Real Leverage
I intentionally chose problems where AI created an "unfair advantage." Here is how I applied intelligence across 2025:
1. Product Execution
- LocalShare: P2P local file sharing. Focused on removing friction from the PRD to a high-impact landing page.
- Notelz: AI-powered aesthetic note generation for students. Iterated until the UI felt "worthy."
- Inforvi: Continuous app-level experimentation, using Google AI Studio for image analysis pipelines.
2. Specialized Intelligence
- LensAI / Lovable: Fashion identification and location recognition via image-based inference.
- ORLON.OG: Location discovery using OSINT-inspired pattern recognition.
3. Automation & Distribution
I deliberately avoided high-effort personal branding. Using Hera AI, I built faceless, voiceless, aesthetic geolocation content.
Consistency scales when humans are removed from the loop. Automation allowed me to focus on building, not performing.
Systems Over Motivation
I do not rely on discipline alone. I rely on Leverage Multipliers:
- Trackers & Checklists: If it's not in the system, it doesn't exist.
- Long-Horizon Planning: Targeting a 10 GPA in BTech CSE (AI/ML) and preparing for higher studies in Japan (JLPT N5).
- Architecture over Tickets: AI did not just make me faster; it made me more architectural. I now think less in tasks and more in systems that produce outcomes.
For Recruiters Reading This
If you are looking for someone who:
- Ships instead of postures
- Thinks in systems, not tickets
- Uses AI responsibly and effectively
Then we will likely work well together. This retrospective is about being predictable under complexity.
Final Reflection
AI did not change my direction. It amplified it. I am still doing what I always did: Learn by building, reflect through systems, and compound quietly.
The tools will change. The operating principles will not.
2025 Core Stack:
- Next.js (App Router) & Bun
- Gemini 2.0 Flash & GPT-4o
- Image Intelligence (OSINT / Pattern Recognition)
- Systems Thinking (PRD-First Workflow)