Blake Hyland
Software engineer with Elixir at the core and a recent focus on AI systems in production. Work ranges from optimizing complex queries to building LLM pipelines that process millions of records on AWS, with evaluation loops that score output quality and retrieval systems over internal document stores. Strongest on backend systems where correctness and cost both matter.
Work History
Full Stack Engineer
6 Months- Built the backend authorization service for a proposal management platform, resolving division-scoped roles so a user could hold different permissions per division, and returning a single allow/deny decision instead of a full permission set.
- Integrated the React frontend with the new authorization API so the interface surfaced only the actions a user's role and team membership allowed.
- Migrated the file subsystem onto the authorization model, auditing every path documents were reachable through so file visibility followed proposal team membership.
- Cut load times on the proposal board for admins in divisions holding hundreds of proposals, where the widest permission scope meant the heaviest query, by paginating the board and list views.
Backend Software Engineer & AI Engineer
3 Years 8 Months- Built an LLM pipeline on AWS SQS and Lambda generating character-limited ad copy for millions of hotel addresses, routing only records over the limit through a model and writing idempotently so redelivery never corrupted data.
- Built an LLM-as-judge evaluation harness scoring pipeline output against explicit criteria, driving prompt iteration until quality met marketing sign-off.
- Cut average query time over 30% across user-facing endpoints through indexing, join restructuring, and spatial indexing with bounding box pre-filters on regional search.
- Eliminated N+1 fan-out in Absinthe GraphQL by introducing Dataloader to batch resolution across a social itinerary feed where users follow travelers and load their published plans.
- Shipped a Slack RAG assistant over ~60 Notion documents using Pinecone, letting employees self-serve HR and policy questions and reducing repeat load on a single-person HR function.
- Led a hotel image overhaul across thousands of properties, automating deduplication and cover image selection with a cascade that shortlisted candidates via Rekognition labels and Python scoring before a single LLM call.
Education
B.S. Accounting, Master of Accountancy
Skills
Certification
MIT Professional Education | Applied Data Science Program, 12 weeks | In progress, expected Dec 2026
Gauntlet AI (formerly Aitra AI) | 8-week AI engineering intensive
University of Utah | Full-Stack Web Development Certification
Timeline
Full Stack Engineer
Backend Software Engineer & AI Engineer
Brigham Young University
Education
Brigham Young University | B.S. and Master of Accountancy
MIT Professional Education | Applied Data Science Program, 12 weeks | In progress, expected Dec 2026
Gauntlet AI (formerly Aitra AI) | 8-week AI engineering intensive: LLMs, prompt engineering, MCP servers, LangChain, Pinecone
University of Utah | Full-Stack Web Development Certification