👋 Hi, I'm Jimi Oso
🎓 Rising Senior @ MIT
📚 Major: Electrical Engineering & Computer Science (EECS)
📈 Minor: Economics
🌍 From: Nigeria 🇳🇬 | Based in: Cambridge, MA
I'm a software engineer and researcher building at the intersection of ML infrastructure, retrieval/evaluation systems, and distributed backend systems. I like turning fuzzy, hard-to-measure problems — is this search result actually good? is this embedding model actually better? — into concrete, testable pipelines.
Currently, I'm a Machine Learning Engineering Intern at Expedia Group, where I built Cross-EG Bench, an internal embedding-model evaluation framework used across Expedia's lines of business for reproducible retrieval benchmarking. I'm also an Undergraduate Researcher at the MIT Media Lab / Latent Lab, prototyping AI-native file-system concepts and building an LLM-judge evaluation pipeline for retrieval tasks.
Previously, I built a hosting-recommendation platform used by 20+ Massachusetts state agencies as a Software Engineering Intern at EOTSS, and was a Software Engineering Fellow at Jane Street's IN FOCUS Fellowship.
Languages
Python Go C/C++ TypeScript JavaScript Java SQL R MATLAB RISC-V Assembly
Frameworks & Tools
FastAPI Docker Kubernetes Google Cloud Platform AWS React Next.js Git WebSockets
Libraries
NumPy pandas TensorFlow PyTorch LangChain scikit-learn Plotly Matplotlib Seaborn
Databases
PostgreSQL Redis MongoDB Firebase
- TrueSplit – Group expense-splitting engine on a double-entry ledger with idempotent Go APIs; Redis + PostgreSQL enforce exactly-once effects under concurrent duplicate requests; Next.js dashboard with live balance updates over Server-Sent Events
- SentinelStream – Real-time fraud detection pipeline in Go using Redis Streams consumer groups and five explainable rule-based signals, including GeoHash-based impossible-travel detection
- Cross-EG Bench (Expedia Group) – Internal embedding-model evaluation framework enabling reproducible retrieval benchmarking on LLM-judged and human-annotated datasets
- Go deeper into ML infrastructure, retrieval/evaluation systems, and distributed backend systems
- Build tools that make it easy to measure whether AI systems are actually working, not just whether they run
- Grow a network of curious builders, researchers, and technologists
- 📂 Resume
- ✉️ Email: jimixoso@mit.edu
"Code should be as elegant as the ideas behind it."