Iker Huang
Toronto, ON · iker.huang@mail.utoronto.ca · github.com/Lunear01 · linkedin.com/in/iker-huang · ikerhuang.dev
Summary
Third-year Computer Science Specialist at the University of Toronto who ships LLM products real people use daily: tool-calling agents, retrieval-grounded assistants, and multimodal document extraction in Python and TypeScript on the Anthropic, OpenAI, and Gemini APIs. Focused on the engineering around the model — tool schema design, grounding, evaluation, and cost. Seeking an AI/ML Engineering internship.
Education
University of Toronto, Toronto, ON
Expected May 2028
Bachelor of Science, Computer Science Specialist — International Merit Admission Award
Coursework: Data Structures and Algorithms, Software Design, Object-Oriented Programming, Systems Programming (C), Linear Algebra, Discrete Mathematics
Experience
AI Engineer (Contract), Import and Logistics Client, Remote
Jul 2026 – Present
- Built an LLM bookkeeping assistant in daily production use: plain-language messages in three languages and currencies become structured ledger entries via schema-constrained output.
- Added multimodal receipt extraction (image and PDF to structured fields), routing inference across Gemini, DeepSeek, and OpenAI endpoints to trade off cost against reliability.
- Designed a human-in-the-loop confirmation step so no model-proposed entry commits without staff approval — the core safety control for an agent writing to real financial records.
- Gate every deploy on a Vitest suite scoring extraction accuracy against fixture cases; built on Cloudflare Workers, Hono, and TypeScript with D1 and R2.
Founder and AI Platform Engineer, Sinapsis, Remote
Jun 2026 – Present
- Architected an AI assistant over a self-hosted ERP that acts through a whitelisted tool schema, grounded strictly in tool results — it reports "not found" instead of inventing records.
- Scoped execution so every tool call runs under the end user's own authenticated session, making guardrails a property of the architecture rather than the prompt.
- Gate all writes behind a field-level confirmation dialog and block protected fields before render; ground procedural answers in a curated, version-correct corpus.
Co-Founder and Technical Lead, TheoryY, Remote
May 2024 – Present
- Built an equity-research agent in Python that drafts analyst memos grounded in live SEC EDGAR filings and FRED macro series, with citation enforcement so no figure ships without a source.
- Developed MCP servers and agent skill definitions exposing financial-data tools to LLM clients.
- Own stack selection, deployment, and production operations for every AI product the studio ships.
Projects
Tuxie's Wiki, Project Lead
May 2024 – Present
- Lead a community documentation site publishing 20+ technical guides for non-expert readers on VuePress and Node.js; 10+ GitHub stars.
- Coordinate outside contributors through GitHub issue tracking and pull-request review.
SkyPath, Developer
Apr 2025 – Sep 2025
- Graph and tree-based route optimization in Python; benchmarked pathfinding approaches on runtime and correctness.
Skills
AI and LLM Engineering: LLM application development, tool calling, AI agents, agentic workflows, prompt engineering, retrieval-augmented generation, structured output and JSON schema validation, hallucination mitigation, human-in-the-loop confirmation, multimodal extraction (OCR), evaluation, MCP
APIs and Frameworks: Anthropic Claude API, OpenAI API, Google Gemini API, DeepSeek, LangChain, Claude Agent SDK, REST APIs
Languages: Python, TypeScript, JavaScript, Java, C, SQL, Bash
Data and Backend: SQLite and Cloudflare D1, relational schema design, database migrations, data pipelines, third-party data APIs (SEC EDGAR, FRED), caching, rate limiting
Cloud and Tooling: Cloudflare Workers, R2, Docker, Docker Compose, CI/CD, Linux administration, Git, GitHub, code review, Vitest
Spoken: English (fluent), Mandarin Chinese (native), Spanish (professional working proficiency)
Awards
UofTCTF 2026 — 15th place, UofT Division · PicoCTF 2026 — 100+ challenges solved · International Merit Admission Award