Gebril Fradj

Incoming Software Engineer Intern (May 2026) on the Windows + Devices (E+D) organization, contributing to core platform and device-facing software used by millions of users worldwide.

• Assisting in improving AI model reliability through structured data validation and QA workflows in the Handshake AI Fellowship.

• Built a scalable photogrammetry pipeline in Python that automates the conversion of 10,000+ 2D specimen images into high-fidelity 3D models, enabling public API access to previously siloed scientific data. • Designed a fault-tolerant, batch-processing architecture to handle large-scale image ingestion, preprocessing, and reconstruction, reducing manual curation time by over 90%. • Developed internal tooling and data validation workflows to ensure geometric accuracy and metadata integrity across thousands of reconstructed biological specimens.

• Architected a secure, AI-driven CI/CD pipeline using Python and Jenkins, accelerating code review cycles by 40% by automatically delivering LLM-generated feedback into Bitbucket pull requests. • Engineered a custom REST API wrapper to integrate multiple LLMs (GPT-4o, LACI), reducing feedback generation costs by 60% through optimized token chunking strategies for enhanced code security. • Containerized the full application with Docker and deployed it via a secure Jenkins pipeline, establishing a scalable system that was adopted for pilot use across four engineering teams.

• Co-authored a UF Journal of Undergraduate Research paper (Spring ’25) benchmarking Whisper, M2M100, and Coqui for offline English,Spanish, and French speech-to-speech translation, achieving under 1.5 s end-to-end latency on a single RTX 3060. • Released the fully reproducible pipeline on GitHub, earned a $1,750 AI Scholar stipend, and presented the work to a 40-person faculty panel

• Modernized the Asia-Pacific IT request funnel, cutting average ticket resolution from 4 days to 2 days (-50%) for 11,000 users. • Deployed a Hugging Face and ML.NET microservice that auto-tags knowledge-base documents, boosting retrieval precision 30% across 500+ weekly queries and eliminating manual work. • Shipped a full-stack submission UI, RBAC dashboard, and email-alert flow; managed Azure DevOps CI/CD releases across 3 regions with a 100% unit-test pass rate. • Led intern team on request-funnel rebuild; prioritized backlog, delegated UI and API tasks, and resolved blockers to deliver the submission feature two weeks ahead of schedule

• Co-authored and published a peer-reviewed paper benchmarking Whisper, M2M100, and Coqui models, achieving <1.5s end-to-end latency for speech-to-speech translation on a single GPU. • Released the fully reproducible pipeline on GitHub and presented findings to a 40-person faculty panel, securing a $1,750 AI Scholar stipend for research contributions.

GPA: 3.84 / 4.0 Awards & Honors: • UF AI Scholar – $1,750 stipend for faculty-guided research on speech-to-speech translation • Bright Futures Academic Scholar • Horatio Alger Engineering Scholar • Herbert Wertheim College of Engineering Dean’s List Leadership & Involvement: • Technical Officer – Software Engineering Club • Academic Chair – Pi Delta Psi • Member – UF ACM, Society of Software Developers, ColorStack National • Volunteer – Tree Club (beach cleanups), Circle K International (community service), Meadow Park Elementary (K–5 tutoring)