Making Generative AI More Token-Efficient
An experiment exploring how far token consumption can be reduced without compromising answer quality — achieving 30% optimization.
Token OptimizationGenerative AI
AI PRODUCT BUILDER
BuildingProductsthat matter
λ 380–740nm·N-BK7 Prism ClusterIOR 1.52 — Optical Crown Glass
Scroll000%
Pune — 202601 — ABOUT
CS grad building AI-powered products at the intersection of technology, product thinking, and human behavior. I take ambiguous problems from early direction through prototyping, experimentation, and implementation.
Beyond the screens, I’m usually listening to a good business podcast, exploring the universe through astrophysics and space tech, or playing the occasional badminton game.
PRODUCT DISCOVERY
USER-CENTRIC DESIGN
AI-ASSISTED EXECUTION
02
WORK
Projects & Systems
LIVE
Reimagining how people learn, interact with, and assess their understanding of PDF content
02.1
I designed and built a RAG-based learning assistant that turns uploaded PDFs into an interactive study experience, combining text and voice Q&A with summary points, three-level structured notes, quizzes, and a session analysis feature to assess the user’s learning patterns and readiness with the PDF content.
0+
Learning Workflows
0%
Token Optimization
Impact
Turned passive document consumption into an active learning experience, bringing explanation, practice, and self-assessment into a single workflow.
LIVE
Bringing the entire career decision journey into one connected product
02.2
I shaped Sankalp’s product vision and architecture, designing the UX and information architecture across all six modules while developing Margdarshak’s recommendation, chatbot, and roadmap logic.
0
Connected Modules
3rd
Prize at Avishkar Project Competition
Impact
Designed to make career guidance more accessible for students, working professionals, and people looking to upskill or transition into the IT industry.
ONGOING
Reimagining how people discover, build, fund, and participate in the businesses of tomorrow.
02.3
I’m shaping Vistar from the ground up, defining its product vision, architecture, and interconnected venture journey while designing the experience across its core modules. I’m also developing the AI-driven systems that connect opportunity discovery, feasibility, venture formation, participation, and ongoing operations into one product. Vistar is being built for aspiring entrepreneurs, investors, and people looking to participate in ventures beyond traditional paths to business ownership.
0
Interconnected Modules
1st
Prize at Avishkar Project Competition
Open Live Project→
In Development — Not Fully Functional
Independent Researcher
ONGOING
Studying human behavior to build products that help humans understand and trust them.
02.4
I’m studying consumer psychology and the perceptions people form while assessing a product, exploring how those perceptions influence their expectations, understanding, and trust. The goal is to develop a framework for designing products that align with and satisfy the mental expectations they create in their end users.
Research Progression
Phase 01 — Human–AI Trust
Literature synthesis examining calibrated reliance, algorithmic aversion, and cognitive mechanisms in AI interaction.
03
WRITINGS
Selected Notes & Essays
The Economics of Free
How tech companies turn zero-price access into adoption, habit, and revenue.
Product StrategyBusiness Model
04
CAPABILITIES
A selection of my most relevant skills and technologies.
Stack, Methods & Tools
STACK
Technical & AI Foundation
“What I build with”
01Python02RAG03LLMs04LangChain05FAISS / Vector Search06Structured Prompting07Java08C / C++09DBMS10React11TypeScript12FastAPI13PostgreSQL14PostGIS
METHODS
Product & Systems Thinking
“How I think and build”
01Modular Product Architecture02Product Ecosystem Thinking03User Journey Design04Product Discovery & Feasibility05Behavioral Psychology in AI06Human-AI Interaction & Trust07AI Context & Token Optimization08DSA & Problem Solving09Object-Oriented Programming10SDLC11Agile / Scrum12AI-Driven Workflows13Location Intelligence
TOOLS
Platforms & Software
“Execution environments”
01Figma02Jira03Git / GitHub04Streamlit05Flask06Android Studio07Hadoop / Hive08Claude Code · Codex · Antigravity09Vercel10Render11Supabase
05
RECOGNITION
Certifications & Honors
AI Product Management Professional Certificate
Coursera · IBM
Project Management Fundamentals
IBM
A
3rd Place — College Level
Avishkar Project Competition
Selected for Zonal Level
German — A1
Goethe-Institut
Generative AI: Foundation Models and Platforms
Coursera
06
Contact // Open to Opportunities
Get in Touch
2026 Atharva More · All Rights Reserved