Ankur Lahiri

Data Scientist

Bengaluru, Karnataka, India11 yrs 9 mos experience

Most Likely To SwitchHighly Stable

Key Highlights

Delivered over $500M in cost savings at Walmart.
Led a team to optimize supply chain operations.
Developed advanced ML models for real-time predictions.

Stackforce AI infers this person is a Supply Chain Data Scientist with expertise in Machine Learning and Optimization.

Contact

Skills

Core Skills

Data ScienceMachine LearningForecastingAnalyticsRisk AnalysisFinancial Modeling

Other Skills

Anomaly DetectionArtificial Neural NetworksBERTBaggingCloud ComputingClustering AlgorithmsDashboard developmentData AnalysisData MiningDeep LearningEnsemble modelsExploratory Data AnalysisFactor AnalysisGenerative AIKPI forecasting

About

Staff Data Scientist | Walmart | Supply Chain & Last Mile Delivery | Forecasting & Optimization Experienced Data Scientist with over a decade of expertise in developing large-scale Machine Learning, Statistical Modeling, and Optimization solutions to tackle complex supply chain challenges. At Walmart, I’ve spearheaded ML architectures and route optimization frameworks that have delivered over $500M in cost savings across transportation operations in 4 years. My work spans global supply chain forecasting, asset utilization, and advanced Last Mile Delivery solutions, including driver-matching algorithms and fraud risk models that enhanced efficiency and reliability. As a Technical Lead, I manage a team driving KPI forecasting and operational optimization across the organization. Proficient in big-data platforms (Hadoop, BigQuery, Teradata), scalable ML model development (Python, R, PySpark), and building automated workflows integrated into real-time analytics dashboards (R-Shiny, Tableau). Passionate about leveraging data to drive impactful business results. Let’s connect: ankur.lahiri91@gmail.com.

Experience

11 yrs 9 mos

Total Experience

3 yrs 11 mos

Average Tenure

5 yrs 11 mos

Current Experience

Walmart global tech india

3 roles

Staff Data Scientist

Promoted

Dec 2023 – Present · 2 yrs 5 mos

Senior Data Scientist

Promoted

Oct 2021 – Dec 2023 · 2 yrs 2 mos

Working as a Data Science Technical Lead in the Supply Chain Transportation Domain in Walmart US.
Responsible for building end-to-end Enterprise Level Pipelines using Data Science and Rest APIs around effective planning of workload, capacity, and assets to ensure smooth operations in transportation in order to guarantee in-stock and on-time delivery for omnichannel customers. The success of this planning is heavily dependent on the availability of reliable forecasts of key workload indicators & asset needs, as well as accurate predictions of demand at different nodes and lanes of the Supply Chain.
Transportation Logistics, before I joined Walmart, consisted of a lot of manual, excel & gut-based decisions with basic models in play. Post that in the last 2 years, my team have transformed some major areas with deployment of ensemble models to generate accurate predictions of workload & assets for entire US Supply Chain network for various time-frames. This solution leverages DC demand & purchase orders to generate the expected workload & capacity forecasts with over 94% accuracy and is delivered on a dashboard. Resulted in $35-50M savings in successive years.
The vision in Transportation is to build a complete holistic optimisation solution over such ML forecasts. This will result in efficient driver staffing & inventory of Fleet, reduction in excess tractor & trailer spends along with better driver reallocation & increased business productivity & error-reduction due to removal of manual processes. Could easily touch savings north of 100M$.
Managerial responsibilities involve leading a team of 3 Data Scientists & 2 interns. Also an integral Part of Walmart Hiring Drives for Data Scientists, Mentoring Data Scientists & Interns to navigate the Supply Chain data & domain & Collaborating with external vendors to finish projects parallelly & derive more value.
Participant in multiple Hackathons - first runner-up Prize, 5th place & semi-final qualification.

Data ScienceMachine LearningForecastingOptimizationKPI forecastingOperational optimization+2

Data Scientist

Jun 2020 – Oct 2021 · 1 yr 4 mos

Predict the real-time ETA of trailer arrival at the store for US Supply Chain - Real-time modeling architectures for efficient prediction of the unplanned break patterns of the trailer drivers in their Outbound journeys from DC to stores, and deployed it in production. ~Annual $2M+ savings due to store-dwell reduction along with an increase in driver productivity, associate workload optimization & improved labor utilization.
Transportation Workload & Assets Forecasting & Optimisation across US Supply Chain - Workload Forecast for Inbound & Outbound Loads for US network across DCs, CCs, stores & vendors, along with route optimization, asset forecasting & driver staffing metrics. ML forecast & Optimization algorithms led to efficient planning & transportation systems for DC, CC & Store delivery. ~Estimated $39M+ savings.

Real-time modelingForecastingOptimizationData AnalysisData Science

Healthlucid

Principal Data Scientist

Nov 2015 – Jun 2020 · 4 yrs 7 mos · Bangalore

Recruited as the first Data Scientist, Healthlucid is an innovative healthcare start-up transforming US health care using leading-edge analytics & big data. Healthlucid addresses the core challenges of the US healthcare system: member satisfaction and total costs.
My responsibilities were primarily revolving around exploratory data analysis, risk & quality control measures for models, understand business use-cases & develop frameworks from existing resources, implementation & last mile execution to commercialize the product.
Regularized regression models for selection automation of medical & pharmacy plans by predicting the best possible cost-effective outcomes & expenditure of employees for the upcoming year, given the set of procedures, doctor visits & medication usage. ~650$+ yearly premium payment savings per client.
Iterative rule-driven clustering algorithms are used to build episodes of care spread across primary & ancillary procedures, multiple days considering hospital admits & separate pricing scenarios for hospitals & doctors.
Anomalous claim detection classifier to filter potential fraudulent claims. ~87% modeling accuracy using Gaussian classifiers. ~$1.2M suspicious claims detected across the book of business.
Built a 20,000+ across-country provider database using google geocoding & Yelp APIs. Deduplication & grouping of same providers across insurances using DBScan to ensure maximum coverage & adoption for our app. ~2x increase in potential target clients. Revenue uptick of $3200+ for HealthLucid based on increased coverage for existing clients. Clustering on provider features (commute distance, time, provider rating & reviews, services available) to influence employee behaviors across different providers.

Exploratory Data AnalysisRisk ControlClustering AlgorithmsAnomaly DetectionData ScienceAnalytics

Credit suisse

Quantitative Risk Analyst

Jul 2014 – Oct 2015 · 1 yr 3 mos · Mumbai, Maharashtra, India

Interest Rate Models on Cross Currency Basis Swaps: Modelled a Value at Risk (VaR) project, extensive application on interest rates modeling grounds, validated the model by developing algorithms in R and Visual Basic (VBA), developed pricer that assessed potential risk in terms of US Dollars by using time series data beginning from the stress period.
Interest Rates models on Volatility Ratios: modeled Risk-not-in-VaR (RNiV) model based on the proxy of the underlying trade volatilities with that of the corresponding swaption volatilities. This risk type pertains mainly to bonds and their futures, data was cleaned by using multivariate techniques, developed R codes to test the linearity of risk, which in turn helped in developing suitable pricers for validation.
Credit Spread Model on Recovery Rates: Model-based on Credit Spreads, which deals with the total debt recovered from the obligator in case of default. Priced the model and tested for appropriateness of certain inherent assumptions. Led to implementing the model in Excel.

Risk ModelingStatistical AnalysisTime Series AnalysisRisk AnalysisFinancial Modeling

Central statistical organisation

Data Science Intern

May 2013 – Jun 2013 · 1 mo · Delhi, India

Analysis of Wholesale Price Indices - Computation and Analysis of the Inflation Rates for various Commodities. Computation of Average Yearly Indices. Analysis of the Dispersion and Consistency of Index Numbers & Forecasting.
Prediction of Fertilizer Demand of Khandwa, Morena, Madhya Pradesh - Data fitted using suitable regression polynomial. The errors are calculated and tested for significance. Then one step ahead prediction is done by the regression polynomial. Data Modeled using time series modeling. Trend estimated by the method of differencing, Seasonal Components by Ratio-to-trend, and moving averages method. Then errors are estimated using ARMA Model assuming the Causality of the process. Then Seasonal Holts-Winter is applied for future prediction. Neural Network Model fitted to the data and for predicting one step ahead.
Time Series Analysis on Rainfall Data - Trend Component estimated by the method of polynomial fitting and differencing. Seasonal and random fluctuations are estimated by fitting a suitable ARIMA model. Predicted several steps ahead using Seasonal Holts-Winter Modelling of the data.

Data AnalysisTime Series Modeling