01
AI for Education (RAG Systems)
Textbook chunking → embeddings → vector search → dual-language answer generation.
Diramu Godana
Computer Science & Economics at Harvard
Undergraduate at Harvard studying CS and Economics. Building bilingual AI tutors, healthcare software, and data-driven systems for social impact.
Get in TouchWork Experience
Harvard College GenAI Research Program
Software Engineering Intern
Jun 2025 — Aug 2025- Built a bilingual RAG tutor enabling Swahili-speaking students to access curriculum-aligned content.
- Processed 1K+ textbook chunks with LangChain, Pinecone, and GPT-4 Turbo, achieving 95% retrieval accuracy.
- Designed RESTful APIs and optimized PostgreSQL schemas in FastAPI for chapter- and question-level queries.
- Developed three tutor modes (summarization, revision Q&A, free-form) with a Next.js/Tailwind frontend and FastAPI backend, achieving <2s response times.
- Piloted with 56 students; dual-language (English + Swahili) output improved comprehension and aligned with syllabus.
Kerugoya County Referral Hospital — Kerugoya, Kenya
Software Engineering Intern / Priscilla Chan Fellow
Jun 2024 — Aug 2024- Built a hospital management portal automating records, billing, and scheduling for 1,000+ patients, cutting wait times by 30%.
- Developed ERP/EMR modules in Java/Python with Frappe, deployed via Docker, and integrated MySQL + REST APIs.
- Worked across Agile sprints on feature design, reviews, debugging, and testing; reduced post-deployment defects by 25%.
Harvard Undergraduate Machine Intelligence Community (HUMIC)
Project Lead & Web Scraping Engineer
Sep 2024 — Present- Promoted from Fellow to Project Lead; developed Harvard’s first self-learning assistant using CrewAI multi-agent frameworks.
- Built web scraping pipelines with Python (BeautifulSoup, Requests) and BraveSearch API to collect educational content.
- Coordinated profiling, ranking, and roadmap teams integrating user data for personalized learning.
Harvard WECode
Associate Director of Engineering
Sep 2025 — Present- Co-develop WECode’s inaugural engineering portal for the world’s largest student-run women-in-tech conference (1,200+ attendees).
- Implement authentication, registration, and event logistics using React and Supabase.
- Oversee technical planning, code reviews, and deployment cycles ensuring scalability and maintainability.
Nyeri Hospice — Nyeri, Kenya
Software Development Intern
Jul 2019 — Aug 2019- Built a DeskAlert messaging system in Python, Node.js, and MySQL to coordinate staff-patient communication.
- Implemented message queuing and logging to improve reliability and reduce response times by 40%.
Featured Projects
Adaptive Swahili-English AI Curriculum Tutor
AI-powered bilingual tutor for Form 1 Biology, built with LangChain, Pinecone, GPT-4 Turbo, and FastAPI. Supports summarization, revision Q&A, and free-form modes with dual-language (Swahili + English) output.
WECode Conference Platform
Full-stack platform for authentication, registration, and mentorship matching. Built with React, Supabase, and CI/CD pipelines for reliability and scalability.
AI-Powered Multilingual Flashcard Generator with TTS
Interactive bilingual learning app that generates English and Swahili flashcards with GPT-4 summaries, TTS audio, and a pastel-themed interface.
Nairobi After — Immersive Web Narrative
Scroll-based digital story built with HTML, CSS, and JS. Features glitch animations, ambient soundscapes, and dynamic chapter transitions.
About Me
I'm a Harvard student studying Computer Science and Economics, passionate about AI for education, data systems, and socially grounded computing. My work spans AI tutoring systems, healthcare automation, and digital storytelling. I've interned at Kerugoya County Referral Hospital, built a bilingual AI tutor through the Harvard GenAI Research Program, and led HUMIC's multi-agent self-learning assistant. I also serve as Associate Director of Engineering for Harvard WECode and as Secretary of the Harvard African Students Association.
My Areas of Specialization
02
Backend Engineering
FastAPI, PostgreSQL, REST APIs, Docker, Pinecone vector stores.
03
Frontend Design
Next.js, React, Tailwind — clean pastel-themed UIs for learning tools.
04
Data & Algorithms
Triangle counting, Strassen matrix multiplication, runtime analysis.
05
Multi-Agent Systems
CrewAI agents for retrieval, summarization, and adaptive self-learning workflows.
06
Product & Research
Iterative development, curriculum alignment, and evaluation for real-world impact.