Grace Jackson

Sophomore at UC Berkeley

Oviedo, Florida, United States

About

Feel free to reach out and connect at [email protected]!

Experience

  • Machine learning contractor at Ariat International
    Jun 2026 - Present · 2 mos

    - Owned the full ML lifecycle end to end for a demand forecasting and inventory system, from data cleaning and feature engineering through model development and deployment - Architected a two-model system combining unsupervised clustering with a gradient-boosted forecaster, and developed a novel uncertainty calibration technique to improve prediction reliability on sparse-history products - Built a frontend dashboard for planners to query model outputs directly, plus an ensemble layer blending multiple base models for improved accuracy

  • Co-Founder / Executive Director at Bake it Forward
    May 2021 - Present · 5 yrs 3 mos

    - co-founded this initiative to provide baked goods and treats to the local domestic violence shelter and homeless shelters - baked goods and collaborated with community bakeries and companies to provide $5000 and counting worth of baked goods - reached 250 people so far

  • Student Research Assistant at UC Berkeley Electrical Engineering & Computer Sciences (EECS)
    Jan 2026 - Jul 2026 · 7 mos

    Developing algorithms and solvers for perfect-information games, utilizing combinatorial game theory and data structures(SQL) to map state spaces and determine optimal strategies under Professor Dan Garcia

  • AI Strategy Consultant at Adobe
    Jan 2026 - Apr 2026 · 4 mos

    - Engineered scoring pipeline across 15+ metrics (MMLU-Pro, HHEM/FCS, etc.) to benchmark an AI system; identified inference bottlenecks and model output friction points to improve response quality and efficiency - Automated evaluation framework to process 10K+ model outputs - consulting club project

  • Strategy Consultant at Ariat International
    Sep 2025 - Jan 2026 · 5 mos

    Developed a Monte Carlo model to quantify the marginal cost of service-levels and engineered a markdown engine with Go/No-Go frameworks processing datasets across 500+ SKUs to deliver insights via 3 mediums for varying technical levels