Victor Chowdhury
I build the systems that decide what a business does next: metric layers, experiments, and the AI agents that put insight directly in a leader's hands. Then I help Product and Engineering ship it.
- Walmart
- Sam's Club
- Nike
- Wipro
About
Sixteen years in analytics and AI, including nine-plus at Walmart, spent in the seam between what the data says, what engineering can build, and what the customer actually needs. I currently lead a 25+ person organization across four pillars (Self-Serve/Chatbots/AI, CRM, Workforce Management, and Ops) and have shipped alongside Product teams in customer service, fulfillment, facilities, fuel & convenience, and store innovation.
Impact, by the numbers
Experience
-
Director, Strategic Insights & AIJun 2025 — Present
Agentic AI, conversational AI, and service economics for a 25+ person, four-pillar organization.
-
Director, Advanced AnalyticsWalmart Fulfillment Services · Feb 2024 — Jul 2025
Insight and measurement strategy for a multibillion-dollar fulfillment business: multi-channel, carrier, and consolidation network economics.
-
Director, Data AnalyticsIntelligent Retail Lab · Jul 2022 — Feb 2024
0→1 emerging technology: computer-vision shrink prevention and next-generation checkout evaluation.
-
BI ConsultantNike Workforce Insights Network & Client Engagement · Feb 2014 — Jan 2017
Grew the Nike project team 4x in 18 months. HR analytics, client-facing discovery, and vendor partnerships (Alteryx, Domo) as part of the pre-sales team.
-
BI DeveloperNike & GMACI · Sep 2010 — Feb 2014
Cognos reporting, SAP HR modeling, and mainframe-to-modern analytics migration.
Highlights by area
The same career through the four lenses people usually ask about.
Agentic AI, GenAI, computer vision, and conversational AI, built to change a decision, not just add a feature. My focus is always the measurement framework underneath: how do we know it worked?
AISLE1: Putting Certified Metrics in a Conversation
Built an agentic AI insights product that serves as the org's governed metric repository. Leaders query certified definitions in natural language and get narrative answers with context, not just tables, and the semantic layer keeps the number the same no matter who asks. It absorbed the recurring ad hoc analysis load and now powers “4-in-the-box” product, ops, and engineering reviews.
Sierra Voice AI: Defining What “Good” Looks Like
Owned the measurement framework (containment, resolution, cost per contact) that leadership used to judge the shift from traditional IVR to an AI voice assistant. Self-serve rate rose from 30% to 70%, with CSAT climbing alongside it rather than trading off against it.
Agent Assist: From Pilot to Scale
Partnered directly with Product on AI-based voice and chat agents on the CRM platform: requirements and problem framing, pilot design, launch readiness, adoption tracking, and post-launch iteration. Built sentiment analysis into the measurement layer to track how customers actually felt about AI-handled interactions turn by turn, not just whether the contact resolved. That caught tone and satisfaction drift early enough to fix it before it showed up in CSAT. Part of a broader CRM AI push that delivered $36M in realized savings.
Computer Vision at the Store of the Future
Directed the data analytics and ML team behind a computer-vision shrink-prevention platform. Behavior and shrink-pattern analysis drove material product improvements, preventing $330M in loss and producing a filed patent (Virtual Cart Optimization Tool).
I build the analytics that decide what gets built: instrumentation, funnels, experiment design, and the success metrics a product is judged on. I work embedded with Product from problem definition through launch and scale.
Shipping the Instrumentation Behind the Roadmap
Own product analytics for the full service portfolio: instrumentation, journey funnels, containment and escalation analysis, and self-serve deep dives that tell Product exactly where customers dropped off and which experiences to rebuild first.
Walmart+ Fuel Benefits: Adoption from Day One
Partnered with the Walmart+ product team to launch a member fuel benefit, building the adoption and redemption tracking that connected fuel behavior to membership growth and retention from the first day of launch.
Smart Fuel Pricing Engine: Closing the Competitive Gap in Real Time
Designed and built Fuel & Convenience's first automated pricing tool in 2020: continuous WTI/Brent crude and Kalibrate competitor-pricing feeds, a configurable radius and price-gap threshold, and a BI layer that quantified the dollar cost of any unaddressed breach before alerting the pricing team via Power BI and mobile. Replacing manual, once-a-day price checks with 24/7 monitoring drove an 8.5% quarter-over-quarter increase in gallons sold.
Tech Assist: Turning Maintenance Reactive to Proactive
Partnered with Product to deploy live refrigeration and HVAC telemetry monitoring, giving field teams a signal before equipment failed and turning a maintenance queue into a proactive product experience.
Metric Governance as a Product Decision
Established definitions, architecture review, and sign-off processes that gave Product one trusted source of truth for roadmap and prioritization, replacing dashboard sprawl with an actual decision system.
I work at the seam where physical footprint, operating model, cost structure, and customer experience meet, framing the right problems and converting complex operational challenges into scalable capabilities.
Building Fuel & Convenience from the Ground Up
Built the strategy and data foundation for a new division from zero, standing up the metrics an entire business needed to operate, and drove a 26% merchandise sales lift by connecting fuel traffic to inside-store purchases.
Middle-Mile Strategy: Internal Consolidation Centers
Helped shape the build-out and operating case for Internal Consolidation Centers, modeling flow, capacity, and cost-to-serve to inform how volume moved through the network, part of a 35% GMV lift for the fulfillment business.
One-Page Decision Briefs for 200+ Leaders
Translated network economics (cost to serve, capacity, carrier performance) into a one-page format used every week across Weekly Business Reviews, making complex trade-offs legible fast for a multibillion-dollar business.
From Reactive Tickets to Proactive Asset Strategy
Built the data and IoT foundation for Facilities Maintenance (architecture, telemetry flows, asset hierarchy), shifting the operating model from ticket response to prevention and informing capital versus expense decisions on repair-or-replace.
Enterprise data only creates value if people trust it and can actually get to it. I build the metric repositories, governance, and self-service layers that make that true at scale.
A 300K+ User Analytics Platform
Owned capacity, row-level security, and self-service enablement for an enterprise analytics platform serving 300,000+ users, and designed a data-lake landscape that shifted the org toward self-service.
NLP Self-Service at Sam's Club
Shipped natural-language self-service so senior leaders could query their own data in free-form text, then trained them on it, treating adoption as seriously as the technical build.
Experimentation as a Practice, Not a One-Off
Ran control-group testing and IVR modernization measurement, built AHT trackers, and recovered IEX forecasting accuracy from red to green, turning experimentation into a standing capability rather than a special project.
Text Analytics to Find the Root Cause
Used text analytics to diagnose root causes across the bottom 10% of Sam's Club locations and target intervention precisely, grounded by running field inspections as first-hand user research.
Call Complexity Scorecard: Scoring Every Contact
Built a weighted, six-factor model that scores every contact Low, Medium, or High complexity from contact reason, workflow count, talk %, repeat calls, transfers, and genuine/non-genuine intent, giving CES a defensible, auditable score to strategize call routing, headcount and skills planning, and scheduling around the real complexity mix instead of raw contact volume.
Work & writing
Live, interactive artifacts and write-ups. Demos use illustrative data to protect Walmart's.
Fruit Slicer & Math: A Game I Built So My Son Would Practice Math
My son, who is on the spectrum, plays exactly one game: Fruit Ninja. It had started pushing ads and a $30-a-year subscription button he kept tapping by accident, while my wife struggled to get him to sit down for math he's actually great at. So I built my own version: a calmer, slower slicing game that pauses every 10 seconds for one multiple-choice math problem and one spoken grade 2-3 spelling word he types in. He gets one retry before the problem changes, and when he ends the session his mom gets a report card showing first- versus second-attempt results. No ads, no paywall, and the screen time now counts as practice.
Agent Performance Scorecard
A month-by-month scorecard that grades every agent on a weighted composite of up to 10 KPIs across quality, productivity and adherence: CSAT, contacts per hour, AHT, shift and break conformance, shrinkage, call quality and transfer rate. Each KPI is scored against a benchmark, with AHT personalized to the agent's channel mix, and an org-wide lens sits beside a department-relative one. Growth-oriented tiers map to a coaching action, and fairness gates keep low-volume, ramp-up and support months out of the grade. Demo runs on fully synthetic data.
Smart Woffie: An AI Insights Agent That Doesn't Make Things Up
An AI analytics agent for Walmart's customer care org that writes SQL against billions of rows in BigQuery, builds executive-ready reports, and keeps its knowledge in a database instead of a prompt, so any analyst on any machine loads the same 'brain.' New findings are saved as drafts and only become trusted once a human validates them, and a two-layer anti-hallucination design (schema rules plus prompt rules) means it only states facts it can cite, with confidence scaled to how well it knows each domain.
Promo & Appeasement Analysis Dashboard
Multi-tab analysis of promo code issuance, redemption, and GMV lift by category and membership tier. It's the kind of decision-ready insight behind the appeasement strategy work that lifted utilization from 51% to 73.5%.
CES Weekly Update: Searchable, Not Just Skimmable
Reimagined the standard weekly status report as a searchable, taggable archive instead of a deck nobody reopens. Search surfaces matches across every past week instantly, and each update carries built-in Appreciate / Feature Request / Escalate actions. A routine reporting ritual became something leadership actually engages with.
Call Complexity Scorecard
A weighted, six-factor scoring model that classifies every contact as Low, Medium, or High complexity from contact reason, workflow count, talk %, repeat calls, transfers, and genuine/non-genuine intent, turning a vague sense of 'this queue feels hard' into a defensible, auditable score leadership can route, coach, and staff against.
Call Complexity vs. Self-Serve Rate
Pairs the Call Complexity score against Self-Serve Rate by workflow and over time. A strong negative correlation shows self-service winning on the easy stuff while the residual, agent-handled mix quietly gets harder, the kind of insight that keeps an SSR win from masking a rising AHT problem.
AISLEONE CRM Portable Dashboard Demos
Six sanitized, standalone reporting dashboards (Agent Assist, AHT Savings, AHT Deepdive, Fixit Ops, QSR, and Spark), rebuilt as zero-backend, single-file demos with synthetic data so the full product analytics story (AHT trends, SLA compliance, contact-reason mix, cost per order) can be shared and explored without exposing production data.
Creative Solutions: The Store Summary
When live Tableau/Power BI dashboards kept failing store leaders on bad in-store WiFi and unpredictable BI servers, the fix wasn't a better dashboard. It was removing the live connection entirely. A leader emails a store number, Power Automate filters the data and renders a PDF + open-work-order list, and both land back in their inbox in about 120 seconds, readable offline before they ever walk in the door.
From Data to Story: The Weekly Narrative
Before AI copilots existed, a ~30-page weekly scorecard export got hand-translated into a single leadership email every week, with every KPI framed three ways (week-over-week, year-over-year, vs. plan), backed by a formal root-cause process, and closed out with a curated scan of industry and competitive news so leadership always knew whether a miss was Walmart-specific or market-wide.
Intelligent Retail Lab: From Faster Checkout to Smarter Shrink
Two computer-vision workstreams, one shared discipline: for the next-gen self-checkout unit, I broke down thousands of reviewed transactions into scan-time and tender-time micro-phases to pinpoint exactly which technology and design choices actually cut friction; for asset protection, I built the forecasting model that turned raw shrink-detection data into a risk-adjusted, defensible savings and forecast number for leadership.
Creative Solutions: The Smart Fuel Pricing Engine
Designed and built in 2020, this was Fuel & Convenience's first automated pricing tool: it ingested WTI and Brent crude benchmarks alongside Kalibrate's third-party competitor pricing, let the pricing team set a radius and a price-gap threshold, and ran 24/7 to flag any competitor breach with the site, the delta, and an estimated dollar cost of not repricing, delivered via Power BI and mobile alerts.
Claude vs. Power BI/Tableau: A Working Model, Not a Cage Match
A framework for where each tool actually earns its place: governed dashboarding for ‘what is happening’ versus reasoning-driven, narrative-ready insight for ‘what does it mean, and what should a leader do about it.’ Includes an illustrative cost model showing why the two tools’ economics scale on completely different axes (viewers versus reasoning work).
Stop Being Overwhelmed. Start Delivering What Matters.
A six-step field guide for analytics teams drowning in ad hoc requests. Six steps (Intake, Classify, Question, Score, Validate, Launch) turn a chaotic request queue into a defensible, confidence-backed prioritization process instead of whoever-asks-loudest.
Education
Licenses & certifications
9 credentialsClick any card to verify with the issuer. Hover to pause.
Certificate in Product Management
Verified credential from Cornell Tech covering product strategy, discovery, and roadmapping, the foundation underneath the product-analytics work in the Experience section above.
Certificate in Machine Learning
Verified credential from Cornell's Ann S. Bowers College of Computing and Information Science: the technical grounding behind the ML deployment and computer-vision work at Intelligent Retail Lab.
15.071x: The Analytics Edge
Verified certificate from MIT Sloan's Operations Research and Statistics faculty, covering data-driven decision-making, regression, classification, and optimization, taught by Dimitris Bertsimas and Allison O'Hair.
6.00.1x: Intro to Computer Science and Programming Using Python
Verified certificate from MIT's Department of Electrical Engineering and Computer Science, the Python foundation underneath the analytics and tooling work throughout this site.
Tableau Desktop Specialist
Certified on essential Tableau Desktop knowledge and data analysis: the dashboarding and visualization skill set behind the BI reporting work referenced across the Product and Data lenses above.
Alteryx Designer Core
Certified on Alteryx Designer's core data-prep and workflow-automation toolset, part of the analytics engineering behind the pricing, reporting, and pipeline work referenced throughout this site.
Core Designer Certificate
Certified on Dataiku's core visual data-prep and workflow-design toolset, another pillar of the analytics-engineering skill set underneath the pipeline and automation work referenced throughout this site.
Google Analytics for Beginners
Certificate of completion from Google Analytics Academy, covering core web analytics and measurement fundamentals underpinning the instrumentation and funnel-analysis work in the Product lens above.
Introduction to Generative AI for Data Analysis
Part of the Microsoft Generative AI for Data Analysis Professional Certificate, covering prompt engineering, AI-assisted data cleaning, and responsible-AI considerations directly underneath the agentic AI work in the AI lens above.
Honors & awards
8 honorsFrom Walmart, Sam's Club, Nike and Wipro. Tap a photo to see it full size.
Peak Captain, Peak '24
Named a Peak Captain for Walmart Fulfillment Services' 2024 holiday peak, when I ran weekly sprint planning with the operations teams and kept analytics focused on what mattered most during the busiest weeks of the year.
Agile Mindset Award
Recognized by Walmart Fuel & Convenience for an agile mindset in 2020, the year I built the division's data strategy from the ground up and launched its first automated, self-alerting competitive pricing tool.
Certificate of Recognition: “Great Safety Dashboard”
Presented by Sam's Club Safety & Compliance leadership for the enterprise safety dashboard, a solution that brought all the relevant safety data into one place so clubs could better manage safety performance.
Certificate of Appreciation
From Nike's HR leadership team for dedication and commitment on the global HR technology program, thanking me for being an integral part of its worldwide go-live.
“Thanks Victor, for all you're doing!”
A pair of Nike boots with a hand-signed note from the Nike team I supported as a Wipro consultant. One of my favorite reminders that client trust is built one delivery at a time.
Thanks a Zillion Team Award
Wipro team award crediting the success of Nike's SAP HR data warehouse project to my hard work and intensity to win.
Feather in My Cap Award
Wipro recognition for the effort and commitment that made Nike's SAP HR go-live happen successfully.
Virtual Cart Optimization Tool (Patent Filed)
Other honors: Certificate of Appreciation · Feather in My Cap (×2) · Thanks a Zillion · IT Wiz 2005
Languages
English (Full Professional) · Hindi (Full Professional) · Bengali (Professional Working)
Recommendations
Received · 5From LinkedIn, shown as received.
If I could pick one person on the planet to work with to create insights through data, it would be Victor. With all sincerity! Happy to talk to anyone that wants more detail.
Victor was a pleasure to have on the team, dynamic, creative and hard working. He brought a thoughtful approach to developing our data strategy and layered on top creative analytics distribution. I wish him well in his current role.
In his one year on my team, Victor created data analytic reporting that had never existed previously on various fronts, even though he had no prior Facilities Services experience. He has always displayed superb professionalism and genuine care for those around him — highly emotionally intelligent and able to navigate uncomfortable situations with grace. Any organization would be lucky to have Victor as a leader on their team.
Victor has always impressed me with his authenticity, compassion and people-first approach to leading teams, and is always informed on the latest developing trends in the BI/Analytics space. Victor has challenged our group to think differently and empowered us with the tools and knowledge to be successful in the workplace.
If you're looking for someone with integrity and a strong work ethic, you'll find that in Victor. Enthusiastic, professional, innovative, and always looking forward and upward, Victor is one of the few people I know that can plow through ambiguity and tough situations with a can-do attitude. I miss having him on the team!