Lead Data Scientist & AI/ML Engineer with 9+ years in analytics and ML, including 3+ years building production GenAI systems at Fortune 50 scale. Expertise in LLM fine-tuning, RAG pipelines, multi-agent orchestration, and MLOps and 5x Google Cloud Certified. Delivered $20M+ business impact | Shipped chatbots serving 200+ daily users | Reduced LLM costs by $15k/month | Optimized $50M+ media spend (18% ROI) | Mentored 8 engineers (3 promoted).
- 🧠 Architected and deployed multi-LLM orchestration (Gemini 2.5 Pro/Flash/Lite, Llama) and Neo4j Knowledge-Graph RAG assistants for over 200 analysts.
- 📈 Delivered multimodal AI prototypes (text + vision) and LLM-powered executive reporting, cutting manual reporting efforts by 60%.
- 🛠️ Led the fine-tuning of LLMs using LoRA, QLoRA, and PEFT, and established robust MLOps pipelines with Vertex AI and MLflow.
- 🏗️ Engineered a 40% increase in data processing efficiency by optimizing ETL workflows and integrating over 50TB of data from 15+ disparate sources.
- 🤖 Currently focused on building advanced multi-agent systems with LangGraph, AutoGen, CrewAI, and the Google Agent SDK.
|
|
🏢 Gen AI / ML Engineer, The Home Depot Management Company | Jan 2025 – Present
- Designed and developed scalable generative AI systems using transformer-based architectures (GPT-4, BERT, Gemini, Longformer) for text summarization, Q&A bots, and contract parsing.
- Led fine-tuning of LLMs with LoRA, QLoRA, and PEFT methods using HuggingFace Transformers to improve model alignment with Home Depot-specific customer and vendor datasets.
- Built and deployed image-to-text pipelines integrating Stable Diffusion and Vision Transformers for intelligent product tagging and visual search enhancements.
- Created custom prompt optimization frameworks and integrated GenAI tooling for internal analytics automation, reducing turnaround time by 55%.
- Implemented RLHF to train chat-based support agents and automated documentation tools, increasing task resolution accuracy by 38%.
- Established MLOps pipelines with TensorFlow Extended (TFX), Vertex AI Pipelines, and MLflow for model versioning, validation, and deployment in staging and production environments.
- Partnered with engineering, product, and marketing teams to align model outcomes with business KPIs through dashboards and experiment tracking.
- Developed a modular GenAI microservice architecture on GCP using Docker, Cloud Run, and Firestore to power multiple realtime analytics and automations.
- Published internal technical documentation and model cards to guide ethics review, reuse, and reliability tracking for all deployed GenAI solutions.
🏢 Senior Data Scientist - Decision Analytics | The Home Depot Management Company | Jun 2023 – Jan 2025
- Architected a robust, scalable dynamic image generation pipeline utilizing state-of-the-art vision models (Google Image Gen3, Stable Diffusion) and advanced text generation models (Gemini-1.5 Pro, Text-Bison-32k), transforming Home Depot’s guided search with enhanced visual relevance and accuracy.
- Integrated multiple AI models, including multi-modal and text embeddings (Text-embedding-004, Gecko@002), to automate image-keyword alignment, optimizing search coherence and improving product discovery across Home Depot’s platform.
- Deployed LLM-powered automation for end-to-end process reporting, delivering scalable, on-demand PDF documentation with code explanations for technical and non-technical stakeholders, cutting manual reporting effort by 60%.
- Developed predictive models including image classification, object detection, and house renovation score prediction using Res-Net and Vision Transformers on MLS listing images provided by CoreLogic, achieving an accuracy of 87% and aiming to save $20 million in marketing budget.
- Led a team of 4 offshore resources (TCS) to optimize ETL workflows and develop ETL scripts in Analytical Workbench and Big Query, integrating data (~50TB) from 15 disparate sources (clickstream, orders, marketing data), resulting in a 40% increase in data processing.
- Built OLAP data cubes and architected databases, data warehouses to support Tableau Dashboards (SAIM Deck) and advanced data analysis, leveraging SQL optimization, clustering, partitioning, stored procedures, and functions on Google Cloud Big Query.
- Partnered with the Data Engineering team to migrate 25+ workflows from AWB Workbench to Google Data form, targeting significant process efficiencies by automating Big Query SQL workflows, dependency management, and job scheduling—expected to cut project timelines by 30-50% upon completion.
- Advancing data pipeline reliability and scalability through Data form’s integrated testing, GIT version control, and optimized SQL transformations, with anticipated benefits of a 50% reduction in data error rates and enhanced handling of large datasets for improved operational efficiency.
- Empowered BACE (Brand Advocates and CEX) partners with advanced analytics tools, enabling real-time monitoring and post-event analysis during critical events such as Black Friday and Cyber Monday to track key metrics, improving strategic decision-making for various projects to enhance customer experience.
- Created data standards and implemented new methods of capturing tagging information in Adobe Analytics Tag Manager by working with the Adobe Analytics team (AAPES team) to gain new analytical insights on customer interaction across all Home Depot online platforms.
- Enabled 5 internal organizations to devise strategies for performing full category refreshes across all Home Depot online platforms to maintain and improve foundational stability of online categories (display taxonomy) and perform full-funnel analysis and optimization.
🏢 Senior Data Analyst | The Home Depot Management Company | March 2022 – Jun 2023
- Analyzed customer behavior across Home Depot platforms to provide key insights to the Category Experience team (CEX) and Brand Advocate Team (BA) with over 300 associates by providing ad-hoc data, standardized real-time reporting, and offering business recommendations for senior executives.
- Enhanced the full-funnel customer experience by providing insights into online Category Pages, Product Listing Pages (PLP), and Product Information Pages (PIP).
- Constructed analytical dashboards using visualization tools like Tableau and Google Data Studio. Leveraged job orchestration tools such as Analytical Workbench and performed data manipulation using Big Query and Python (~70 hours/month).
- Delivered website performance analytics using Adobe Analytics, Tableau, and Big Query to derive analysis of 15+ events (Black Friday, Winter Sale, etc.) over the year to improve the Click-Through Rate (CTR) and Conversion Rate (CVR) by optimizing content placement.
- Led a team of 7 in the Voice of Associates (VOA) initiative, leveraging the Liftoff platform to streamline data science onboarding, reducing onboarding time by 20% and increasing satisfaction by 10% through department-specific insights shared with senior leadership.
- Acted as an Adobe Analytics Workspace SME, leading weekly Adobe Analytics Office Hours to provide live training and creating training resources and best practices.
- Fostered cross-functional partnerships, mentored junior analysts, delivered technical training and spearheaded various project initiatives.
🏢 Data Analyst and Engineer | Harley Davidson Motor Company | Feb 2020 – March 2022
- Performed in-depth analysis of general merchandise data to identify opportunities and develop proposals and recommendations for use by management.
- Designed, prepared, and manipulated data using Business Intelligence tools—Tableau, Power BI, and SAP Analytics Cloud—to identify user behavior and analyze trends and patterns, both independently and in collaboration with product managers and data modeling resources.
- Extracted, cleaned, and analyzed multiple data sources, and built optimized data models and ETL pipelines to support dashboard requirements using SQL and Alteryx, which improved the performance of existing reporting dashboards in SAP Analytics Cloud and helped reduce data processing time by 80%.
- Maintained, enhanced, and drove Root Cause Analysis in conjunction with SMEs to identify and resolve business process problems, leading to a decrease in open purchase orders by 55% and inventory count mismatches by 30%.
- Maintained the master dataset of the General Merchandise department and performed batch inserting/updating of accounts, product information, lead times, BOM, dealer information, and other objects in SAP using FLEX PLM.
- Worked with the Supply Chain Analyst and warehouse coordinators to perform error analysis on EDI transactions (IDoc Resolution), providing recommendations and analyzing all error data established for new product builds and launches, compiling and communicating weekly metrics to leadership.
- Led multiple rounds of User Acceptance Testing (UAT) by identifying appropriate stakeholders and building test scripts for each to execute.
🏢 Marketing Analyst | Anahata Art and Design Pvt | May 2019 – Dec 2019
- Created, maintained, and managed a 3-week Google Ads Campaign with a total budget of $300 for an online gifting startup in India to understand their business, market competitors, popular selling products, and target audience.
- Managed to make 110 ad copies with 6,000 keywords, minimized cost-per-click to $0.11, achieved a 200% increase in website traffic (92% new Users/week), achieved 113 sales of products with $3100 in revenue, and improved the landing page experience.
- Proposed multiple recommendations for the potential new markets, product updates, and marketing campaign changes to the client after analyzing visits, page views, purchases, revenue, and conversion metrics from multiple data sources (web analytics data as well as external data).
🏢 Data Scientist | Principal Financial Group | Aug 2019 – Dec 2019
- Devised the KPI's and predicted market regime of companies from Russell 1000 to evaluate prospective investments for the client.
- Performed data aggregation on 5.5 million rows of data using SQL aggregation techniques.
- Implemented data cleaning and hyperparameter tuning using machine learning algorithms in Python that resulted in an accuracy of 78%.
- Increased the prediction accuracy of investing in one company by ~5% by identifying the most important variables that increased the client's confidence in carrying out deals/investments.
- Tools and Libraries- Python, R, Microsoft SQL Server, Logistic Regression, Random Forest, XGBoost, dplyr, scipy.io, keras, numpy, matplotlib, sklearn.
🏢 Graduate Assistant | University of Maryland | May 2019 – Dec 2019
- Assessed, created and maintained student records (~4000 students) to identify low performing students and scheduled sessions to improve academic standing and disciplinary records.
- Performed data extraction using SQL to assist the academic advisors to implement various programs to improve a student's performance.
- Led a team of 10 undergraduate students to answer student inquiries that improved the satisfaction rate by 10%.
- Designed visualizations to understand and analyzed the career opportunities chosen by the students and helped advisors to improve the academic program based on the visualizations.
- Tools- SQL, SIS(Student Information System), Excel.
🏢 Data Analyst | Bridge Solutions | May 2017 – May 2018
- Created visually impactful and interactive dashboards in Tableau and Excel to report various key KPIs of various clients of Bridge Solutions.
- Handled and built relational databases, designed queries using Microsoft SQL Server, and created reports for analyzing and root-causing board failure data. Well-versed in finding patterns and trends in complex, multivariable data sets using Python in an agile environment.
- Created inventory targets by employing analytical abilities, data mining skills, and experience that resulted in a cost reduction of $1M.
- Introduced and developed Docker application to deploy IBM OMS 9.5 and WMS 9.5 which is used by 75% of the workforce.
Master of Science in Business Analytics (MSBA) Coursework: Big Data and Artificial Intelligence, Data Analysis with Python, Data Mining and Predictive Analytics, Data Models and Decision Making, Database Management System, Operation Analytics, Decision Analytics, Price Optimization and Revenue Management, Google Analytics
Bachelor of Technology in Computer Science Coursework: Web Technology, Software Engineering, Pervasive computing, Operating Systems, Network Security, Microprocessor and Interfacing, Linux Internals, Database Management Systems, Data Structures and Algorithm Design, Data Mining, Artificial Intelligence and Expert Systems
Technologies: Tableau Visualizations, Dashboards, MS SQL Server, Excel, Wix
- Built a relational database by scrapping online data of apartments near College Park, MD.
- Performed data cleaning, ETL operations, populated and normalized the database to 3NF using SQL queries.
- Developed interactive dashboards using Tableau to help users compare the price, location, amenities for various houses and published the same to a website developed using Wix.
- Gained 2000 users in a span of 1 month and converted 250 users to active leads for the apartments.
Technologies: Machine Learning, Text Mining, Predictive Analysis, NLP, Python
- Implemented a model using Amazon reviews data (10 GB) to predict two most similar products based on item-item collaborative filtering using Python.
- Cleaned, trained, cross-validated, performed NLP operations like tokenization, lemmatization, stop word removal on the dataset and used KNN model to achieve highest accuracy of 84.4%.
- Built visualizations to analyze the reviews and built word clusters to discover the most used words for each review level using matplotlib and seaborn.
Technologies: Machine Learning, Python, Predictive Analysis, Neural Networks, Transfer Learning
- Built a machine learning (CNN) model to predict the age and gender of a person via their photos using python.
- Acquired 7 GB of data from IMDB-WIKI to perform data cleaning, data aggregation and data manipulation.
- Implemented transfer learning from VGG-face model, tuned the model to prevent overfitting and implemented the model on real time video stream.
- Achieved a mean absolute error rate ±4 years in predicting the age and an accuracy of 97% in predicting the gender.
Technologies: RStudio, Machine Learning, SQL, Predictive Analysis, Visualization, R
- Analyzed a 60K+ dataset to predict patient revisit rate for emergency care hospital patients and discovered the correlation with other features.
- Performed exploratory data analysis, multiple imputation and feature engineering using R.
- Implemented and evaluated models like Logistic, KNN, Ensemble models, Decision Trees, XGboost.
- Secured 2nd position by achieving ~83% accuracy, thereby reducing the patient readmission rate by 4%.
- Planned and evaluated leadership growth programs, utilizing a fund of $80,000 per semester.
- Coordinated logistics for numerous events with club presidents under the Smith Master Student Association (SMSA).
- Launched an alumni networking platform to foster relationships and mentorship for incoming and existing students.
- Collaborated with the program manager and director to improve the MSBA program.
- Resolved cohort grievances, achieving a 30% reduction in grievances from previous batches.
- 📧 Email: Send me an Email
- 🔗 LinkedIn: Connect on LinkedIn
- 📱 Phone: (240) 360-7905
- Lost 90 lbs (45kg) in two years! Hit me up for healthy alternatives to Indian food!
- Led a team of 10 undergraduate students as Graduate Assistant at University of Maryland
- Coordinated events and managed a $80,000 fund per semester as Executive Vice President at Robert H. Smith School of Business
- Labeled issue #1 in ajay-sai/VSML_Fine_Tuning
- ❗ Opened issue #1 in ajay-sai/VSML_Fine_Tuning