Kiran Vijay

1. 01Scout2026 · AIAn AI data analyst you can check. Ask a business question in plain English; Scout loads the right skills, writes and validates the SQL, and answers with the query and sources behind it. Runs on a custom agent harness I built from scratch, over fictional sample data.Next.js · TypeScript · SQLite · OpenAITry it↗

2. Try Scout · ask it a question

   Scout · sample answerQ1 FY27 (Jul–Sep 2026) · fictional data

   ExpenseRevenueHeadcount

   Why is Cloud over plan on expense this quarter?

   Where it looked

   Skill

   Cloud expense analysis: matches “Expense vs plan this quarter”

   Tables

   Plan and actuals, Expense detail

   Checks

   4 queries checked before running

   ›First query · FIN_PLAN

   SELECT account,
          SUM(plan_amount)   AS plan_amount,
          SUM(actual_amount) AS actual_amount
   FROM FIN_PLAN
   WHERE account_group = 'Expense'
     AND business = 'Cloud'
     AND period BETWEEN '2026-07' AND '2026-09'
   GROUP BY account
   ORDER BY account
   LIMIT 50

   Actual

   $18.33M

   Plan

   $17.58M

   Variance

   +$752k

   Variance %

   +4.3%

   • Cloud hosting+$793k
   • Data center+$19k
   • Network+$2k
   • Marketing−$9k
   • Salaries and benefits−$49k

   Ask Scout a questionAsk Scout →
3. 02AI analytics at BNY2026 · AIHow my team at BNY lets people ask data questions in plain English and get trusted answers and dashboards back. Agents query a governed semantic layer instead of raw tables, every query is validated before it runs, and evals track accuracy over time.Semantic layer · LLM agents · Text-to-SQL · EvalsRead›
4. 03Buddy2026 · ProductA personal life OS for Android that brings tasks, habits, spending, focus and health into one calm daily cockpit.Kotlin · Jetpack Compose · Next.js · FastAPI · SQLiteRead›

Foundations

1. 04Sales Analytics2024 · Power BIRead›

1. 1

   Governed data

   Shape data, definitions, controls, and access so teams can trust the same answer.

2. 2

   Semantic layer

   Semantic models and metric definitions, so people and agents read the numbers the same way.

3. 3

   Data agent

   Takes a question in plain language, writes the query and shows its working.

4. ✓

   Explained answer

   The answer comes back with its sources, so people can check it.

• AI-native analytics

  Data agents, RAG, and prompt-driven dashboards that show their working, so people can ask in plain language and check the answer.

• Trusted data layers

  Curated datasets, semantic models, and metric definitions that give people and agents the same governed answer.

• Data engineering

  SQL, Snowflake, Fabric, and warehouse patterns that make pipelines inspectable, repeatable, and ready for AI.

• Interface craft

  Visual hierarchy, type, and interaction patterns shaped for people who need the answer quickly.