Problem
Selecting high-caliber candidates for the ACE Program required reviewing complex multi-week coding tasks and GitHub repositories. The manual review process was exhausting our engineering staff, creating an unsustainable operational bottleneck and slowing down cohort admissions.
My Role
I led the product direction and system implementation. I framed the operational problem, designed the intake and evaluation architecture, and built ACE_Forge as a specialized platform to automate candidate evaluation.
What I Did
- Built an onboarding portal where applicants receive module-specific engineering challenges based on their current technical track, paired with automatic reminder cadences before the weekly deadline.
- Engineered automated GitHub repository ingestion to fetch code submissions and commit histories directly via the GitHub API.
- Implemented an automated AI grading pipeline that analyzes code submissions, calculates performance scores, and generates structured feedback highlighting strengths and implementation gaps.
- Created an administrative dashboard allowing human assessors to review synthesized AI evaluations, inspect flagged gaps, and fast-track top performers directly to the ACE_Evaluation Meeting.
Tools
Next.js, Node.js, GitHub API, LLM Evaluation Pipeline, Task Scheduling.
Outcome
- Reduced staff onboarding operational effort by approximately half (~50%).
- Eliminated manual triage, allowing technical mentors to focus solely on high-value candidate interviews and specific flagged edge cases.
- Established consistent, objective evaluation standards across all candidate cohorts.