Work
Selected projects in analytics, AI, and decision systems
A mix of enterprise-scale analytics work at Amazon and independently built AI products — reflecting how I think about data, systems, and user impact.
Multi-Agent AI System
Soul Spark
A multi-agent AI life intelligence system that integrates finance, health, career, and growth signals into a unified conversational advisor.
Problem
Most personal data systems are fragmented. Health, money, tasks, and behavior all live in separate apps, making it difficult to turn data into clear decisions.
What it does
Soul Spark integrates finance, health, career, and growth signals into a unified conversational advisor that helps users act on what matters most.
System Design
- Local Ollama LLM inference for privacy-first usage
- Shared memory and action loop across agents
- Packet-native multi-agent architecture
- Integrated signals from structured and semi-structured sources
UX Approach
- Prompt chip interface for quick interaction
- Context-aware advisor responses
- Action-oriented guidance
- Live demo experience
Why it matters
It reflects my interest in moving beyond analytics dashboards toward systems that actually help users make decisions. The shift from "here is your data" to "here is what you should do" is where AI creates real leverage.
Problem
Job seekers spend enormous effort tailoring applications without knowing how well their resume actually matches a role. ATS systems reject strong candidates silently.
What it does
An AI-assisted tool that parses resumes, scores role fit (0–100 weighted algorithm), identifies ATS keyword gaps, and generates tailored cover letters. It surfaces what's missing, what's strong, and what to emphasize for a specific JD.
Why it matters
I built this to solve a real friction point — people apply to roles with no signal on fit. Combining NLP, structured parsing, and LLM reasoning created a tool with genuine utility.
Problem
Choosing the best way to travel between two points is surprisingly complex — car vs. train vs. flight trade-offs involve cost, time, CO₂, and personal priorities that no single tool handles well.
What it does
A route and travel-decision tool that compares car, train, bus, and flight across cost, time, and CO₂. Includes multi-stop optimization, an AI travel advisor with multi-turn chat, and AI-generated packing checklists via local LLM.
Why it matters
The core challenge was integrating multiple data sources with different schemas and presenting coherent comparison logic. The AI layer adds judgment where rules alone fall short.
Problem
Competitive Dota 2 players produce enormous amounts of match data but have limited tools to generate strategic insight — most dashboards show stats, not decisions.
What it does
A match intelligence tool that combines public OpenDota API data with AI-generated strategic insights, draft simulation, and match breakdowns. No authentication required.
Why it matters
A personal project and useful test case for combining structured game data with LLM reasoning. The domain is narrow enough to validate insight quality rigorously.
Analytics systems, data products, and infrastructure built during 9 years at Amazon — spanning payments, private brands, and global marketplace expansion across India, Luxembourg, and multiple regions.
Selection Expansion Analytics Platform
Amazon · 2024 – 2026Amazon · Luxembourg · L6 · 17+ global stores
Problem
17+ global Amazon stores had fragmented, untrustworthy selection data. There was no reliable single view of what was expanding, what the funnel looked like, or how to frame it for leadership planning cycles.
What I built
- Redesigned the entire metric taxonomy — Platform GMS, offer breakdown, Contribution to 3P sales, selection quality — and rebuilt the QBR framework adopted across 4 senior leadership planning cycles.
- Rebuilt core selection identification logic, fixing long-standing data inaccuracies and restoring trust in 20+ downstream reports used daily by Business Managers and engineering teams.
- Delivered entitlement calculations for a $2.5B Seller growth Big Bet under strict timelines, forming the analytical foundation for 2026 annual operational planning.
Private Brands BI Platform
Amazon · 2021 – 2023Amazon · Luxembourg · L5→L6 · EU/NA/APAC
Problem
EU Private Brands had no analytics function. Product, Marketing, Supply Chain, and Finance were all making decisions without reliable shared data, tooling, or a common metrics language across regions.
What I built
- First BIE hire for EU Private Brands — built and led the BI function (3 BIEs + interns), owning the full analytics stack across four business domains.
- Delivered the APB-first Deals Performance Dashboard: 6 views, 20+ metrics, 90+ users across 10 countries and 12 job families. Global single source of truth, saving 350+ hours/year.
- Built Catalog Defects and products-with-a-sale bridge analysis uncovering ~€110M GMS opportunity — adopted by senior leadership for quarterly planning.
- Designed and led Inventory Tracker Engine with global supply chain teams, saving 1,000+ hours/year for EU Private Brands; extended to NA and APAC in V2.
- SQL cube innovation reduced the weekly business review codebase from 20K to 1.3K lines — adopted org-wide across APB BI.
Amazon Pay & HFC Analytics System
Amazon · 2017 – 2021Amazon · Bengaluru, India · L4→L5 · Payments & Commerce
Problem
Amazon Pay and High-Frequency Commerce (Recharges, Bill Payments, Flights) needed analytics infrastructure built from scratch — for product launches, marketing campaigns, fraud prevention, and Go/No-Go experiment decisions, all simultaneously.
What I built
- Built end-to-end data pipelines, 5 dashboards with 15+ views, and a 25+ query self-serve library supporting the launch of 8 HFC categories — foundational analytics for a $2B business driving 50+ launch campaigns.
- Designed the Downstream Impact methodology for product launch decisions — Go/No-Go framework for 5 Pay experiments at $30M GMV stakes, adopted as the PM standard for feature launches.
- Built a Python-based self-serve customer segmentation tool processing ~100 marketing campaigns/month, reducing campaign setup time by ~70% and saving ~3,000 hours/year.
- Designed abuse-prevention rules combining multiple data sources to auto-cancel ~100K abusive HFC orders/month (~2% of volume), significantly reducing incentive fraud with minimal false positives.