Software Development Intern · Evergreen AI
Oct 2025 – Present
On a team of three developers, building and hardening the platform students use to write wellbeing dialogues for LLM fine-tuning.
- Collaborate on a team of 3 developers to maintain a site used by over 200 students to create dialogues for fine-tuning LLMs focused on student wellbeing.
- Prototyped a SetFit few-shot classification pipeline that tagged 13 semantic attributes across 5,000+ dialogues at 93% accuracy, then used the label distributions to rebalance the training corpus.
- Designed a red-teaming sandbox in Flask with an OpenAI-compatible API for testers to probe a candidate chatbot’s safety guardrails.
- Built a round-robin bug reporting and assignment system integrating Slack slash commands, GitHub Projects, and AWS Lambda, cutting average bug resolution time from 3 days to 1.5 days.
- Python
- Django
- Flask
- SetFit
- AWS Lambda
- Slack API