Data Science · AI Engineering

Sahil Devulapalli

Building AI pipelines and data systems that turn raw signal into decisions.

B.S. Data Science candidate at the University of Michigan who shipped enterprise AI applications at AThingz this past summer. I work across the stack — from anomaly detection models to the dashboards that make them usable.

Email me → LinkedIn GitHub

Ann Arbor, MI / 248-525-0417

Experience

Where I've worked

AThingz Southfield, MI
June 2026 – Aug. 2026

AI & Data Engineering Intern

Developed two enterprise AI systems using Python, Django, Azure AI Foundry, Azure OpenAI, Microsoft Graph API, SharePoint API, Azure SQL Database, Azure Blob Storage, Pandas, and Parquet, reducing daily manual review time by 50% across supply chain and logistics operations.
Designed an autonomous email-to-data pipeline that extracted, validated, transformed, and stored structured shipment data using LLMs, processing 120+ shipments over 3 months, then built a full-stack Django shipment management platform with secure authentication, analytics dashboards, and customer self-service shipment entry that increased total shipment volume by 18%.
Translated supply-chain requirements into deployable data and decision systems, delivering validated solutions for intelligent shipment processing and multi-objective constraint optimization across 2,300+ real-world shipments.

Projects

Things I've built

Multi-Objective Dock Scheduling & Re-Optimization Engine

Modeled plant operations as a constraint-programming system generating truck–door–time assignments from multi-table SQL data, enforcing ownership, compatibility, operating windows, durations, closures, and zero overlap.
Built optimization infrastructure with five-level lexicographic objectives, dependency-graph neighborhood expansion, and incremental re-optimization.
Optimized 2,376 trucks across 513K+ candidate assignments, verified with 136 automated tests.

Constraint Programming SQL Optimization Python

LLM-Powered Shipment Data Engineering Platform

Built an autonomous email-to-data pipeline that processed 120+ shipments over 3 months, parsing LLM responses into strict nested schemas.
Persisted validated records with full audit trails to cloud storage; migrated prototype SQLite workflows to enterprise SQL with normalized tables and automated ETL scheduling.
Launched customer self-service shipment entry through a Django app, increasing total shipment volume by 18%.

LLM ETL SQL Django

Cyber Threat Detection using Machine Learning

Implemented anomaly detection pipelines using Isolation Forests and PCA to identify suspicious behavior in 1M+ network log events, achieving 92% recall on attack patterns.
Engineered temporal features including rolling event counts and anomaly rates to improve detection sensitivity.
Built a Streamlit dashboard to visualize risk scores, anomaly trends, and alert thresholds for rapid incident triage.

Isolation Forests PCA Streamlit Python

Personal Finance Tracker

Processed and standardized 5,000+ nonprofit financial transactions into clean, structured datasets with normalized categories and timestamps for anomaly detection and spending insights.
Implemented ARIMA and Prophet models to forecast monthly expenses, reducing prediction error by 22%.
Designed dashboards using monthly spend timelines, category-wise variance plots, and anomaly flags to identify outliers and long-term budget risk.

ARIMA Prophet Data Cleaning Forecasting

Leadership

Outside the classroom

Michigan Sports Analytics Society

Data Analyst · Ann Arbor, MI · 2024 – Present

Trained predictive models using Python (scikit-learn, Pandas, NumPy) on player and team performance data, improving prediction accuracy for team statistics by 17% and generating actionable insights for strategy and player evaluation across multiple seasons.

Collaborated with peers to build visualizations in Tableau and Matplotlib, communicating insights to audiences.

Michigan AI Strategy Club

Vice President · Ann Arbor, MI · 2024 – Present

Lead initiatives to explore real-world applications of AI across industries including finance, healthcare, and logistics.

Organize workshops and events to help members build practical AI skills and stay updated on emerging trends.

Investing4Youth

Co-President, Co-Founder · Novi, MI · 2020 – Present

Co-founded a nonprofit promoting educational equity; raised $3,500+ by analyzing donor outreach data, extracting actionable insights, and optimizing fundraising strategy for maximum impact.

Built the organization's website with analytics dashboards, increasing traffic 40% and securing 5+ partners and 10+ fundraisers.

Skills

Toolkit

Programming & Software Development

Python Java C++ SQL MATLAB Django OOP Data Structures Algorithms Full-Stack Development Backend Development REST APIs API Integration JSON Authentication Git GitHub

AI, Data & Cloud Technologies

Azure AI Foundry Azure OpenAI Prompt Engineering LLM Applications Pandas NumPy scikit-learn PyTorch Azure SQL Database Azure Blob Storage Microsoft Graph API SharePoint API Parquet ETL Pipelines Data Engineering Database Design Data Validation Tableau Power BI

Education

Academic background

University of Michigan — College of Literature, Science & Arts

B.S. in Data Science, Minor in Computer Science · Ann Arbor, MI · Aug. 2024 – May 2028

Certified in Cybersecurity (CC) – (ISC)², July 2026.
AWS Certification (Pursuing) – Expected May 2027.

Cybersecurity Knowledge: Security Principles, Business Continuity, Disaster Recovery & Incident Response, Access Controls, Network Security, Security Operations.

Relevant Coursework: Data Structures & Algorithms, Discrete Mathematics, Probability and Statistics, Database Management Systems, Software/Hardware/IoT, Linear Algebra, Calculus III, Programming & Data Science.

GPA
3.66 / 4.0

Off the Clock