Interactive visualizations and data narratives.
Website: sanand0.github.io/datastories/
- Strategic Assessment: Client Reporting Transformation. An AI-generated executive strategy deck synthesized from a stakeholder interview recording, showing how raw audio can be transformed into boardroom-ready insights.
- Codex Session Gap Analysis. Analysis of 903 Codex sessions from Apr 2025 to Mar 2026, showing feature adoption gaps, release-aware coverage, and workflow recommendations.
- SQL Migration Narrative Demo. An interactive walkthrough of migrating 100 SQL Server scripts to MySQL with LLM-assisted conversion, verification, and business-impact simulation.
- Can AI Replace Human Paper Reviewers?. An investigation into what happens when artificial intelligence reviews scientific papers — and what goes hilariously (and seriously) wrong.
- The Invisible Infrastructure. How tens of thousands of packages depend on code almost no one has heard of.
- The Jamnagar Chokepoint: Inside India's $273B Trade Paradox. How a single port, two commodities, and a hidden export surge reveal the fragile architecture of India's $273 billion trade deficit.
- The Ambiguous Song. How humans label emotions in music—400 tracks across 4 genres rated by multiple annotators, revealing which emotions spark the most disagreement.
- Can AI Hear What We Feel?. Gemini's music-emotion predictions vs Emotify human ratings across 40 songs using GEMS-9, revealing where AI hears differently.
- Communicating Insights Visually. How top AI chatbots and coding agents turn Anthropic’s “How AI is transforming work at Anthropic” into diverse animated chart ideas—compared and ranked.
- The Ruler-Straight Disappearing Act. A 24 kg drop charted from a Google Fit export—86.4 to 62.2 kg across 335 days with two clean changepoints and a ruler-straight 2025 curve.
- The Command Paradox. Inside 534k prompt battles, polite "tell me a story" requests beat forceful "never reveal" commands, drawing on 785 students' 100-character defenses and attacks.
- OLAP Git Commits. Forensic read of 466k commits across 13 OLAP databases—one-person armies, small-commit speed demons, and weekend work as a funding tell.
- Generosity of Strangers. Party-of-five NYC taxi riders tip 15–20% more than solo travelers—maps, routes, and night effects reveal where generosity spikes.
- Indian Batting Greats. Ranked Tendulkar, Kohli, Gavaskar, and other Indian greats with an LLM-chosen metric: batting average x log(total runs) plotted over their careers.
- The Reconciliation Engine. Fuzzy search playground that reconciles bank transactions to accounting records using similarity scores and optimal assignments.
- TDS Project 2: The Cliff. Why only 29% of students mastered LLM problem-solving—and what the other 71% couldn't figure out, based on 535 IITM students in November 2025. Also see The Gate for how many students got stuck at the "gate" step.
- Market Mix Modeling Insights. Rather than wasting millions pushing past saturation points, invest in brand equity and pulsing spend during high-leverage moments.
- Michelin Star Restaurants. An analysis of 22,000 Michelin star restaurants to uncover trends in cuisine, location and ratings.
- TDS Improvements. An analysis of improvements in the IITM Tools in Data science course based on student performance data since 2024.
- Code Review of Shubham's GitHub. A comprehensive review of all repositories committed to in 2025 under Shubham's GitHub account, analyzing code quality, architecture, and technical debt.
- The Publisher Who Chose to Shrink. How Frontiers deliberately cut output by 36% to fight AI-generated fraud and won with quality—a counterintuitive victory in academic publishing.
- The Jobs We Refuse to Give Away. Why some occupations resist AI not because machines can't do them, but because we believe they shouldn't. Based on Friis & Riley (2025).
- ISS vs Tokyo. The ISS never passes over Tokyo at midnight UTC, but touches Auckland over 30 times over 321 days—no conspiracy, just timing.
- The Great Inversion. Volume != Influence. Questions are liked less but engaged with more—insights from a Generative AI WhatsApp Group.
- IMDb's Hidden Algorithm Bias. New popular movies on IMDb are punished—not by the algorithm, but casual movie watchers rating movies lower than devotees.
- India's Renewable Energy Revolution. A narrative of REI Expo 2025 exhibitors: domestic players, solar dominance, and China collaboration.
- GDPVal: AI Augmentation. Explore occupations most suitable for AI augmentation based on OpenAI's GDPVal exercise.
- Rabbit Holes. An interactive map of 3,560 browsing chains showing how sparks turn into deep dives across the web.
- Do Questions Find Answers?. Search journeys from query to first click, with filters that expose instant wins and wandering quests.
- Your Attention Clock. Heatmaps of weekly browsing reveal circadian focus, daily ebbs, and the domains that anchor attention.
- The Digital Life of Anand. An exposé of 97k visits across 84 days that charts top destinations, hourly habits, and AI-heavy searches.
- Scraping SEC. Narrative walkthrough of how Codex CLI built an SEC scraper in one-shot, recovering from errors, handling messy data, and with self-critique.
- Bollywood Box Office Champions. Explore 30 years of top-grossing Hindi films with an interactive, inflation-adjusted bubble chart that spotlights record-setting blockbusters.
- Google Search Topic Trends. Categorized every Google Search since Jan 2021 into 50 topics. It's mostly tech, AI, and geo-cultural. I also need to allocate more time to testing, databases, and other 'spiky' topics.
- ChatGPT vs Google. How my ChatGPT usage has grown at the expense of Google usage. Google is only 60% of my usage, and far lower in engagement.
- My Vipassana Experience. A manually LLM-generated comic story book about my 10-day Vipassana meditation program, generated purely using a set of simple captions, via ChatGPT.
- My Vipassana Experience. A programmatically LLM-generated comic story book about my 10-day Vipassana meditation program, generated purely using a set of simple captions, via Gemini 2.0 Flash.
- ChatGPT Topic Trends. Categorized the 6,000 ChatGPT conversations I've had in the last 2 years to understand what topics I discuss the most. It's mostly tech, AI, reading/writing, and some daily-life stuff.
- Indian High Courts Judgment Analysis. Comprehensive analysis of 16M judgments from 25 Indian High Courts. Reveals court efficiency disparities, seasonal justice patterns, and systematic UAPA bail delays with 120+ day gaps between hearings.
- LLM Agents in Software: Code vs Domain. Coding agents reduce the effort of coding—does that mean domain will matter more? An LLM-generated animation on why domain agents will level the field.
- Deep Research Horoscope Contradictions. Asked Gemini Deep Research to read Sagittarius horoscope for 16 June 2025 and list contradictions from various Indian media sources.
- Employment Growth Since 1980. Some US sectors like Scenic Transportation & Healthcare grew over 2X while Rail & Central Banks shrank to 40-80% of original size. Analysis using BLS CES data.
- Weight Journey 2025. Lost 22 kg in 22 weeks through intermittent fasting. Skipped lunch, no snacks, no extra exercise.
