Vegard is a mathematically inclined software engineer with a strong focus on backend systems, data engineering, and custom algorithm development. They demonstrate advanced proficiency in Go and Python, frequently building low-level systems like interpreters, file watchers, and complex OCR data extraction tools. While highly innovative and capable of solving hard algorithmic problems, their projects currently prioritize exploration and prototyping over production-grade polish and security.
Exceptional use of concurrent I/O patterns to fan out file saving and minimize latency in high-demand endpoints.
Excellent at quickly proving out complex concepts like custom OCR heuristics or database components.
Tendency to write monolithic 'God-objects' (e.g., embedding HTTP, base64 decoding, auth, and database logic inside a single Lambda handler).
Leaves critical security gaps such as unmitigated SSRF vulnerabilities, lack of timeouts, and hardcoded infrastructure configurations.
Dominant language across 8 repositories. Shows advanced usage of concurrency (e.g., errgroup) to build complex systems like interpreters, serverless functions, and porting text editors.
Demonstrated through custom geometric OCR bounding-box mathematics in extract-table and an R package for survival analysis built for a master thesis.
Strong interest in system-level concepts, evidenced by building a SQL interpreter, a macOS file watcher, a redo-logging implementation, and a text editor.
Effectively leverages AWS Lambda, DynamoDB, S3, and Textract to orchestrate complex data extraction, though struggles with connection pooling and security.
Complete lack of test coverage in the flagship extract-table project significantly increases regression risks for complex mathematical logic.