Software Engineer
Chidi Orji
Software engineer with eight years experience across US and UK product companies and an MSc in Artificial Intelligence (University of Salford). Built and maintain Concord — a production expert-matching platform for The College Collective (KWP Ltd), in active use — as the sole engineer from data pipeline through API to React frontend. Known for owning systems end to end: first architectural decision through to deployment, observability, and iteration. Toptal Top 3% vetted talent; currently contracting at Flex (US banking infrastructure) and Autumn AI (acquired by Qualtrics).
Python · SBERT · scikit-learn · TensorFlow · Keras · YOLOv8 / YOLOv9 · FastAPI · PySpark · Apache Spark · MLlib (ALS) · MLflow · SparkSQL · Databricks · Pandas · NumPy · React · Node.js · TypeScript · PostgreSQL
Recognition
Top 3% engineering talent, vetted by Toptal
For a niche dog-training platform, I rebuilt the interface, made it fully responsive, and shipped a sponsorship feature that opened an advertising revenue stream — then restructured the codebase into an Nx monorepo on React 19 and Next.js, consolidating front- and back-end deployment to cut hosting costs. Toptal published the work as a case study.
Read the case study →Selected Work
All projects →- 01
Concord — Expert Matching Platform
Live app ↗Production platform for The College Collective, a UK Further Education peer-support network, replacing a manual spreadsheet workflow. Matches each support request to candidate industry experts across 37 capability areas using a three-signal ranker — SBERT (all-MiniLM-L6-v2) embeddings, TF-IDF cosine similarity, and categorical overlap — served from a FastAPI microservice behind a React / Express / PostgreSQL Nx monorepo. Built and maintained as the sole engineer.
Python · FastAPI · SBERT · TF-IDF · React · TypeScript · ExpressJS · PostgreSQL · Kysely · Nx
- 02
Intelligent Matching System (MSc Dissertation)
MSc Artificial Intelligence dissertation (University of Salford). Research and prototype automating the candidate-identification stage of the College Collective's expert matching. Combines categorical overlap over a 37-label taxonomy, TF-IDF cosine similarity, and Sentence-BERT embeddings into one ranked list, evaluated against 320 historical matching decisions. Achieved Precision@10 of 48.44% and MRR of 0.186, alongside a documented audit of five sequential evaluation-integrity bugs with before/after metrics for each.
Python · Sentence-BERT · SBERT · TF-IDF · scikit-learn · NLP · Pandas
- 03
Field Trial Training Platform
Toptal case study ↗Modernization of a niche canine field-trial training platform: redesigned the UI, added mobile responsiveness, and built a sponsorship feature to open a new revenue stream. Migrated the codebase to an Nx monorepo on React 19 / Next.js, consolidating front- and back-end deployment and cutting hosting costs. Featured as a Toptal case study.
React · NextJS · Node.js · TypeScript · Nx
- 04
Bluelines
Home page ↗Bluelines is a business management solution for SMEs. Bluelines abstracts away the boring paperwork carried out by business owners on a daily basis, allowing them to focus on growing their business.
React · TypeScript · TanStack Query · Python · Django · Graphene
- 05
- 06
Autumn AI
Home page ↗Autumn AI is an early-stage startup based in Canada working on the future of preventative mental healthcare
React · TypeScript · NextJS