⚙️ Production Data Infrastructure · Predictive Analytics · Autonomous AI Systems

I build the infrastructure that drives enterprise intelligence. From idempotent ETL/ELT pipelines and deterministic forecasting models to multi-agent LLM systems optimized for scale, cost, and extreme accuracy.

I operate at the intersection of Data Engineering, Analytics, and Applied AI. I don't just write code; I architect solutions that eliminate bottlenecks and solve high-stakes business problems.

An API-first analytics engine that unifies natural-language querying, AutoML, and one-click reporting into a single, Kubernetes-ready deployment.

Architecture: Engineered a high-throughput FastAPI backend routing to AI query layers and embedded ML microservices (Isolation Forest, Prophet). Infrastructure: Fully containerized via Docker, orchestrated with Helm/Kubernetes, and deployed seamlessly across mixed cloud environments. Business Value: Instantly translates complex CSV/SQL data into board-ready executive PDFs, entirely eliminating the need for ad-hoc manual SQL reporting. Python FastAPI Kubernetes Docker AutoML

An end-to-end Machine Learning pipeline converting raw HR telemetry and call center transcripts into actionable, predictive dashboards.

Architecture: Trained and tuned a Random Forest model on 50,000+ records to predict attrition, deployed K-Means for behavioral segmentation, and executed VADER sentiment analysis on 10,000+ audio transcripts. Business Value: Achieved 91%+ prediction accuracy and integrated the pipeline with Power Automate, slashing weekly stakeholder reporting latency by 60%. Scikit-Learn Pandas SQL Tableau Power Automate

• Mindpex | Agentic AI & Solutions Engineering Intern
  • Architected multi-step LLM workflows and mathematically optimized context windows for a predictive workforce-retention engine. Translated complex HR requirements into deployable AI architectures, drastically reducing token inference costs without sacrificing output intelligence.
• Frost & Sullivan | Data & Analytics Intern
  • Engineered SARIMAX and Prophet time-series models for enterprise market forecasting. Validated high-stakes financial telemetry across 10+ business scenarios using strict RMSE, MAE, and MAPE benchmarking to drive C-level consulting deliverables.
• Infinite Computer Solutions | Software Engineer Intern
  • Built automated, real-time analytics microservices (Node.js/Couchbase) serving 20+ enterprise stakeholders. Replaced 4 hours of weekly manual reporting overhead with a zero-touch, zero-trust ETL architecture.
• Zoho Corporation | DataPrep Summer Intern
  • Executed advanced data profiling, enterprise schema mapping, and pipeline validations to harden core ETL features for Zoho DataPrep 2.0.

1. Idempotency by Design: Re-runs must be safe. Breakage is inevitable; recovery must be instantaneous, programmatic, and strictly side-effect free.
2. Zero-Trust Data Contracts: Validate aggressively at the boundaries. Anomalous data must fail fast rather than silently corrupting downstream machine learning models.
3. Smallest Usable Slice: Ship the thinnest end-to-end path that proves business value, then harden the infrastructure for scale.
4. Zero-Meeting Documentation: Runbooks, architectural diagrams, and explicit naming conventions. The codebase should speak for itself.