Current Work
Automation, AI research, industrial data, archives, robotics, and grid tools.
Automation / AI Research / Software / Data
Yusra Rasool
Focus
Building software and data tools that fill the gap between messy systems, AI ideas, and work people actually need to get done.
Databricks
Azure
SQL
AI
Python
C++
Informatica
Fabric
Power BI
Agents, research, and practical tools.
Pipelines, validation, archives, and dashboards.
Automation systems and interfaces people use.
OT steel data modeled at Nucor
breakout-risk forecast accuracy
event-review accuracy, C3 AI agent
data assets centralized (Informatica)
accounts managed, audit errors
records in the Paperstack archive
Projects
Featured / Industrial Data
Nucor Data Work
Breakout-risk modeling, C3 AI monitoring, and Informatica data marketplace work across Azure and Databricks data sources.
Read experience
Next.js / Cloudflare D1 / NPS + NWS data
National Parks Intelligence
A trip planner that forecasts crowds and weather across U.S. national parks — live NPS and National Weather Service data, when-to-arrive advice, per-park backup plans, and a clickable park map.
Open project
Three.js / WebGL / Elevation rasters
Kashmir 3D Atlas
A free-flowing 3D terrain atlas of Kashmir built from real elevation rasters — fly over the valley around Srinagar with ambient land-cover and travel context layered onto the landscape.
Open prototype
Next.js / D1 / EIA · eGRID · NREL data
AI Grid Pressure Atlas
An ambient U.S. map scoring where AI data centers strain the electric grid — one pressure score per region from EIA, eGRID, NREL, and OSM sources, shown as glowing nodes you can explore.
Open prototypeReact / Next.js / SQL / Cloudflare D1
Paperstack Digital Archive
Built a searchable historical-newspaper archive with React, Next.js, Cloudflare APIs, media previews, source links, reviewer metadata, SQL/D1 schemas, and backfills across 3,000 records.
Open project
Industrial Vision / PyTorch / ResNet-18
BYOL Industrial Defect Detection
Trained Bootstrap Your Own Latent on 12,568 unlabeled steel surface images, then fine-tuned ResNet-18 across four defect categories to show how self-supervised pretraining can reduce annotation effort for surface inspection.
View repoC++ / CUDA / Parallel Processing
C++ Parallel Data Benchmarks
Built CUDA C/C++ kernels for record matching and duplicate-check workloads, using Bloom filters, hashing, memory layout changes, and parallel reductions to make high-volume validation faster and reproducible.
View GitHub
Databricks / PySpark / Lakehouse
Kashmir Tourism Data Pipeline
A Databricks Asset Bundle with PySpark notebooks, Delta Live Tables, API secrets, scheduled refresh jobs, H3 heatmaps, DBSCAN clustering, and validation/deploy commands for repeatable geospatial analytics.
View work
AI-Assisted Systems Tooling
LangGraph Systems Inspector
A multi-agent workflow for inspecting connected data, routing dependency questions, tracing anomaly paths, logging agent decisions, and making repeated investigation runs easier to review and debug.
View repoOpen Source CLI Contribution
Field Theory CLI
Contributing to a JavaScript/TypeScript CLI that syncs X/Twitter bookmarks locally, makes them searchable, and supports agent workflows through terminal access, local storage, and reviewable command behavior.
View repoCurrent Engineering Ramp System
Tech Learning Paths
A website I use to organize learning resources across Windows Server, Visual Studio, SSMS, C#/.NET Core, .NET Windows services, Angular, SQL Server, Azure DevOps/Pipelines, healthcare systems, and AI-assisted coding practice.
Open project
User-Facing Application
Rate My Housing
A Python-based housing review project built around user-facing comparison, practical data capture, and helping people make better housing decisions faster.
View repo
Research Foundation
BYOL on CIFAR
Earlier self-supervised image representation learning work exploring Bootstrap Your Own Latent on CIFAR images before adapting the framing toward industrial inspection and low-label defect review.
View repoExperience
May 2025 - Aug 2025
Data Engineering Intern
Nucor
Industrial data work across Databricks, C3 AI, Informatica, Azure, and governance workflows.
- Developed a Databricks breakout-risk model on 400 GB of steel OT data, forecasting disruptions with 89% accuracy for operational planning.
- Built a C3 AI monitoring agent to flag breakout risk from steel process signals, improving event review with 95% accuracy for faster response.
- Architected an Informatica Data Marketplace to centralize 100+ technology data assets across Azure and Databricks sources for data governance.
Dec 2022 - Present
Undergraduate AI Researcher
TKAI Lab / University of South Florida
AI research in representation learning, training alternatives, and video anomaly detection.
- Defined, implemented, and tested alternatives to standard training approaches, including Local Representation Alignment.
- Worked iteratively through experimental design, implementation, validation, debugging, and comparison under constrained compute settings.
- Documented learnings, assumptions, and results in ways that supported continuity and future experimentation.
- Built a strong research habit around questioning assumptions, comparing alternatives, and making technical findings easier to explain and review.
2025 - Present
MicroMouse Robotics Member
RoboBulls / University of South Florida
Autonomous robotics work in path planning, simulation, and repeatable maze traversal.
- Implemented path-planning routines and simulation tests to validate maze traversal before hardware integration.
- Connected mapping assumptions, route decisions, and robot behavior checks so navigation logic could be reviewed and improved.
May 2023 - Present
Compliance Team Lead
USF Business and Finance
Finance workflow automation, JIRA process design, and compliance-aware operations.
- Maintained responsibility across 156 accounts with zero regulatory failures and zero audit-period errors.
- Led audit and reconciliation workflows through sensitive financial and compliance constraints where data accuracy and confidentiality were hard requirements.
- Used JIRA, process tracking, and Python automation to reduce manual effort, improve repeatability, and tighten validation on invoice and vendor processes.
- Captured process logic and edge cases clearly enough for continuity, verification, and future handoff.
Research & Recognition
AI research, leadership, and honors.
Research & technical notes
Local Representation Alignment
Implemented and tested a replacement training method for backpropagation across autoencoder and variational autoencoder experiments.
DeepMIL Video Anomaly Detection
Built a pipeline to rank abnormal segments in long surveillance videos and reduce frame-level labeling needs.
BYOL Defect Detection
Explored self-supervised learning for steel surface inspection where labeled examples are limited.
Leadership
Society of Competitive Programmers
Led Python workshops, organized qualifiers and review sessions, and helped students build stronger problem-solving routines.
RoboBulls MicroMouse
Collaborated on autonomous maze-solving software, simulation checks, and path-planning logic.
Awards & scholar programs
- IEEE IDEA Scholar2023
- Grace Hopper Celebration Scholar2024
- Tapia Scholar2023 + 2024
- Gold and Green ScholarUSF
Contact