Agentic AI
Conversational analytics agents that answer business questions through deterministic metric tools rather than free-form text-to-SQL, and agentic workflows for everyday engineering.
Data Science · ML Ops · Agentic AI Engineering
Amirtha Ganesh
Pugazhendhi
Seasoned data scientist with 9+ years of experience in statistical modeling, machine learning and analytics, now moving into ML Ops and agentic AI engineering. I build ML models, production data pipelines and AI agents on Google Cloud (Vertex AI, BigQuery, ADK), and design hands-on cloud and ML certification training.
Enterprise & partner experience
Employers, clients and lending partners I have worked with. Logos are trademarks of their owners.
9+years in statistical modeling, ML and analytics
4certifications: Google Cloud PMLE, PDE, ACE + Google AI
2Georgia Tech master's: Analytics (4.0) and Industrial Engineering
1,000+MS Analytics students mentored at Georgia Tech
90+executed hands-on notebooks in a 12-session curriculum
01
Expertise
A data science base, extended into production ML and AI agents.
ML Ops & Cloud ML
Training, pipelines and endpoints on Vertex AI, with reproducible, containerized code and version control from experiment to production.
Data Science & ML
Statistical modeling, forecasting and planning, supervised and unsupervised learning, and deep learning, applied to plans worth billions in sales.
Languages, Data & BI
Production SQL and Python on BigQuery, plus the reporting layer executives read, from pipeline to dashboard.
02
Experience
The Home Depot
Atlanta, GA · 4+ years
Financial Services & Payments (FS&P) Data Science & ML Senior DS
Apr 2025 – Present- Credit sales planning. Built the planning solution (trend decomposition, risk adjustment, promo-lift modeling) that sets store-level credit goals across a multi-billion-dollar sales plan, giving stores clear targets and leaders a more accurate plan.
- Credit applications planning. Run a multi-layer planning engine blending trend, risk, promo and field-adjustment layers into store-week goals for millions of applications and new accounts that drive billions in incremental sales.
- Credit insights agent and insights suite. Leading an AI insights agent (Google ADK, Gemini) so Credit Ops teams can self-serve performance answers, and building a monthly insights suite with ML classification of customer feedback.
- Executive reporting. Own executive and CEO-level reporting end to end; productionalized store-facing credit reporting with EDW and Store Ops IT, including UAT, with no dedicated engineering build.
Operations Data Science DS
Jun 2022 – Mar 2025- Prescriptive ML (Sidekick). Helped build the ML models behind Sidekick, the flagship store-associate app (billion-dollar impact); built an interpretable driver classifier and led SKU-location analysis behind approved purge-logic changes.
- Perfect Bay composite score. Built the bay-quality score with robust statistics for skewed metrics and a sales-impact model; the business adopted its ranking method to prioritize bay improvements.
- Pilot analytics and computer vision. ROI and customer-behavior analysis for cashback and markdown pilots informed rollout decisions; computer-vision inventory insights; embedding-based SKU matching; a PCA tool for store facility labor allocation.
Earlier experience
Kaizen Analytix
Data Science Intern · 8 months
Ensemble-regression forecasting for Toyota Production Systems (forecast error down ~14%); revenue-optimization model for A&E Networks (~$1M savings).
Georgia Institute of Technology
Instructional Associate & Graduate Research Assistant · 2.5+ years
Mentored 1,000+ online MS Analytics students; Daimler assembly simulation research.
Ashok Leyland, India
Deputy Manager, Quality & Reliability R&D · 3 years
Predictive reliability model for commercial vehicles (~$500K savings); multi-criteria supplier selection (~$50K savings).
University of Duisburg-Essen, Germany
Summer Research Intern · May – Jun 2015
Lot-streaming scheduling for flow shops with consistent-size sublots, defective goods and sublot-attached setups, minimizing makespan. Built a mathematical model validated with IBM CPLEX, and a meta-heuristic that reached the optimal solution in fewer iterations and less compute time.
03
AI workshops & hands-on ML engineering
Designed and built to go deep on the full Google Cloud ML stack.
AI workshop series
Designed a 12-session Google Cloud certification curriculum covering ML Engineer, Data Engineer and Cloud Engineer, with speaker-note decks and 90+ executed hands-on notebooks.
ML foundations
A reproducible codebase from classical ML to neural networks, including one CNN implemented in TensorFlow, PyTorch and JAX, with Vertex AI training, endpoints and KFP v2 pipelines; plus Dataflow, BigQuery, IAM and GKE labs.
Academic projects
Cyber-physical attack detection
MATLAB/Simulink system using genetic-algorithm and time-series methods to craft and detect stealthy attacks (~91% accuracy). Won Best Attacker and Best Detector.
H1B visa approval prediction
Compared classification models on historical case attributes, evaluating feature importance and performance trade-offs.
04
Certifications
Each badge links to its public verification page.
05
Education
Georgia Institute of Technology
Atlanta, GA · Aug 2019 – May 2022
- MS AnalyticsGPA 4.0
- MS Industrial EngineeringGPA 3.86
- Management of Technology CertificationGPA 4.0
Anna University, College of Engineering Guindy
Chennai, India · Aug 2012 – Jun 2016
- BE Industrial EngineeringCGPA 9.15/10
- University 2nd Topper, Silver MedalistHonors
Let's connect
Open to conversations about data science, ML Ops and agentic AI engineering.