I build the data layer, and the AI on top of it.

I'm a data engineer finishing an M.S. in Data Science, Analytics, and Engineering at Arizona State University. Before that I spent two years at Celebal Technologies building data platforms on Azure Databricks for clients in healthcare, energy, and manufacturing. Lately I build LLM systems that run on real data, like a private, offline AI system for ASU Surplus. I graduate in December 2026 and I'm available early 2027.

• 2026, in progress · MS capstone, team of four

  Demand Forecasting & Inventory Optimization

  27.7%WAPE vs. 38.5% naive baseline

  Forecasting demand across 250 SKU and warehouse series, with a strict time-based test window. Turned the forecast into a reorder-point policy that cuts simulated holding cost about 28% at comparable service levels.

  Team project with Rakshit Govind Thota, Sanjana Parmareddy Gari, and Yashwanth Sai Tondapu.

  • Python
  • LightGBM
  • Prophet
  • scikit-learn

• October 2026

  End-to-End CI/CD Deployment Pipeline

  Multi-tier AWS environment in Terraform, parameterized so one codebase deploys dev or prod. GitHub Actions runs tests on every push to main, builds a Docker image, pushes to ECR, and triggers a Terraform deploy.

  • Terraform
  • Docker
  • GitHub Actions
  • AWS

  View Repo

• January to May 2026 · Built as a TA for ACO 430 under Prof. Feng Wang

  WLAN Traffic Forensics Toolkit

  100,000+WPA2-encrypted packets parsed

  Rebuilds the network topology as a directed graph and classifies device types with K-Means clustering. A Z-score detector on top caught a live denial-of-service deauthentication attack.

  • Python
  • Pyshark
  • NetworkX
  • scikit-learn

• March 2026

  F1 Race Outcome Predictor

  80%top-3 accuracy vs. 73% baseline

  Ranks drivers by win probability, with rolling features shifted forward to prevent look-ahead leakage. Backtested on 44 held-out races: 52% top-1 and 80% top-3 accuracy, against a 43% / 73% baseline.

  • Python
  • scikit-learn
  • HistGradientBoosting

• September 2026

  Containerized Flask Service on Kubernetes

  Flask API with Postgres and Redis in a multi-stage Dockerfile, run locally through docker-compose and deployed to a local kind cluster.

  • Docker
  • docker-compose
  • Kubernetes (kind)

• 2023

  Real-Time Group Chat Application

  Multi-client chat in Core Java. A server handles concurrent TCP socket connections and broadcasts messages, with client and server code kept modular.

  • Java
  • Swing
  • AWT
  • TCP sockets

1. May 2026 to present

   Tech Assistant, Surplus Operations

   Arizona State University

   Tempe, AZ

   • Built Surplus AI, a private, offline AI system for ASU Surplus.
   • Built a Power BI dashboard on asset sorting data showing where items stall.
   • Advise university departments and outside buyers on what equipment fits their needs.

2. August 2025 to May 2026

   Graduate Teaching Assistant

   Arizona State University

   Network Security & Wireless Networks, Glendale, AZ

   • Designed hands-on labs for two 400-level networking courses covering traffic analysis and packet-level debugging.
   • Held office hours for 40+ students.
   • Built and presented the WLAN forensics toolkit to the class.

3. September 2022 to November 2024

   Associate, Data Engineering

   Celebal Technologies

   Jaipur, India

   A data and AI consultancy. I worked on client engagements across healthcare, energy, and manufacturing, mostly on Azure Databricks.

   • Telehealth client in New Zealand, Microsoft Fabric

     Clinical and operations teams were waiting days for reports. I designed the Lakehouse architecture and coordinated a five-person team. Time-to-insight dropped about 50%, and the dashboards ended up in their weekly operational meetings.

   • A US materials engineering client, AWS

     Ran a production pipeline on Glue, S3, and Redshift handling about 10 million records a day from 7+ source systems. Cut downtime about 30% by tracing recurring failures to root causes instead of patching them.

   • A major US utility, data governance (10 months)

     Their Microsoft Purview catalog was maintained by hand. I built automation on Databricks that pushed entities and custom classifications into Purview on a schedule, plus Apache Atlas lineage pipelines and a business glossary mapped to their governance policies.

   • Medallion architecture on Databricks

     Built bronze, silver, and gold layers plus a serving layer on Azure Databricks and Delta Lake. Wrote the Spark transformations and dbt models.

   • A US medical device manufacturer

     Audited their existing Azure Data Factory pipelines before changing anything, and shipped updates with zero disruption to production.

   • A large FMCG client

     Data governance engineer, using Unity Catalog alongside Microsoft Purview to manage access, lineage, and classification across the Databricks workspace.

   • Production ML pipeline

     Built the Spark and Databricks pipelines feeding a production machine learning model, worked with the data science team on feature engineering, and built the serving layer that delivered model-ready data.

   • A/B testing

     Built the pipelines feeding experiment data, set up test and control assignment logic, and ran the statistical analysis on results.

   • Pre-sales

     Wrote Azure cloud architectures and Bill-of-Materials cost estimates, and presented them directly to prospective clients alongside the sales team.

Contact

Graduating December 2026, available early 2027.

linkedin.com/in/srujan-rudrapati