Developed a document to knowledge-graph pipeline using FastAPI, & relation extraction writing to Neo4j, with Elasticsearch indexing and REST search APIs, doubling relevance and cutting median latency by 50%.

Built a visual-search with: OpenCV preprocessing and torchvision ViT/CNN embedding service behind a FAISS index, exposed via REST; achieved recall@10 0.92 and p95 140 ms at 600 QPS.

Developed a domain-specific ontology to improve catalog search, leading to a 100% improvement in retrieval performance

Designed and deployed an API gateway and 30+ RESTful microservices for Virginia Tech's Electronic Theses & Dissertations (ETD) digital library, leveraging Flask, Docker, and Kubernetes in a large-scale distributed architecture

Integrated LTI 1.3 with OAuth 2.0 and JWT authentication into an open-source platform serving 10,000+ users

Developed a code snapshot tool in Java using Abstract Syntax Tree (AST) analysis to identify student coding patterns, enabling targeted interventions that improved project success rates by 50%.

Implemented automated CI/CD pipelines for an ETD system, reducing deployment time by 60% and improving code quality

Optimized microservices architecture for a Tier 1 bank, improving the scalability by 200% and system performance by 30%.

Developed report analysis and data mapping tools for a multi-tier banking system (Java, SQL), resulting in a 30% improvement in data accuracy and enabling real-time analytics

Created a machine learning–powered ERP forecasting feature to predict financial outcomes based on asset performance, to enable better strategic planning.

Developed an automated web scraping tool with TypeScript (Puppeteer) for Nigeria's power sector, boosting data collection efficiency by 70% for the Ministry of Power.

Developed a COVID-19 infection rate estimator in Flask (Python/Pandas), optimizing algorithms for data-intensive processing to achieve 98% prediction accuracy.

Implemented RESTful APIs using Python and Flask in conjunction with Postgres and MongoDB.

Improved data integrity by resolving MySQL database duplication issues, reducing storage utilization by 30%