Systems Integration Engineer · Boulder, Colorado

Vraj Patel

I turn operational bottlenecks into reliable software and data systems.

At CU Boulder’s Institute of Behavioral Science, I build backend services, automation, operational analytics, and ML-assisted workflows designed for day-to-day use. My research background spans public-interest data and satellite cybersecurity.

Recruiter brief Selected work Resume

Evidence at a glance

82.6% / 0.756: Sealed-test accuracy / macro-F1 for an unreleased BERT routing candidate
10.8K / 96.7%: Current ticket corpus / records carrying support-group IDs
61% / 66% smaller: Raw / gzip analytics payload after replacing Plotly with Recharts
~15 min → <1 min: Typical Active Directory group cleanup after PowerShell automation

How I contribute

Engineering range, backed by specific work

I work across systems integration, backend engineering, automation, and applied ML, with an emphasis on explicit state, review boundaries, and recovery when a long-running job fails.

Backend systems and controlled integrations

My ticket-intelligence work connects source reconciliation, application state, staff review, and approved outbound updates. Model predictions remain separate from authorization.

FastAPI · PostgreSQL · React · Human-in-the-loop ML

See the ticket-intelligence case study

Automation with measured results

I turn repetitive administrative work into bounded tools with clear outcomes, including an Active Directory offboarding workflow that reduced a typical 15-minute cleanup to under one minute.

PowerShell · Active Directory · Endpoint operations

See the experience behind the result

Recoverable data and ML pipelines

I design long-running collection and processing jobs around checkpoints, resumability, and reviewable outputs—from public records to audio and satellite telemetry.

Python · AWS · Selenium · Document processing

See the parliamentary-data case study

Projects

Selected work

Case studies with contribution, constraints, system decisions, and evidence.

All projects

Software Engineering

Operational Ticket Intelligence

Full-stack support operations combining dependable backend workflows, lifecycle-aware analytics, and advisory routing suggestions.

10,803 / 96.7% Current ticket corpus / records carrying support-group IDs

View case study

Research

Indian Parliamentary Data

Multi-vendor speech and translation evaluation followed by a recoverable end-to-end pipeline for parliamentary records, campaign audio, and structured analysis data.

40+ GB 2014 and 2019 campaign audio processed after multi-vendor tool evaluation

View case study

Career