Junhao Zhao
Junior Solutions Engineer at ElcanoTek building AI agents and developer tools for programmatic advertising. I work across backend systems, MCP integrations, reliability, and product experiences in Go and TypeScript.
Tech Stack
Featured Projects
big-brain
Open-source MCP server and CLI that gives every AI assistant one shared memory: your notes, projects, and tasks as plain markdown you own. Published on npm and listed in the official MCP Registry, with 23 tools, local hybrid search, and 287 tests.
Drug Synergy Analyzer
Full-stack research platform for analyzing drug synergy in cancer treatment. A Python/FastAPI backend processes uploaded dose-response datasets and computes synergy scores; a Next.js frontend visualizes results with interactive D3.js and Plotly charts.
Experience
Junior Solutions Engineer
Jun 2026 – PresentElcanoTek
- Deploy and maintain ElcanoTek's product stack for nine client companies, each on its own five-server Vultr environment running single sign-on, the Fleet AI agent platform, an inventory classifier, a client dashboard host, and an email archive, with Cloudflare DNS and an isolated AWS SES/S3 email tenant.
- Scripted client rollouts end to end: creating servers and DNS records behind create-only API guards, registering SSO apps, verifying each deployment, and shipping config updates only while Fleet is idle.
- Built ElcanoTek's central authentication service from the ground up in Go, now the single sign-on every client deployment runs on, with Argon2id passwords, TOTP two-factor authentication, PKCE handoffs, and signed two-way role sync with Fleet, hardened through 20 security-review PRs across four repositories.
- Cut branded-email generation time 4.6–14.6x and model cost 86–94% by replacing model-written HTML with a deterministic renderer validated in an 80-run A/B benchmark, and parallelized MCP connector startup (up to 4.5x faster) with a deterministic tool catalog so the LLM prompt cache keeps hitting.
- Build the data layer behind AI deal agents for programmatic advertising: guard-railed natural-language SQL over multi-billion-row S3 ad logs with AWS Athena at under a cent per typical query, an MCP server for onboarding a new SSP, and performance analysis of a 192K-QPS fraud-detection service running inside an ad exchange.
Undergraduate Researcher
May 2025 – PresentFox Chase Cancer Center / Temple University
- Built a full-stack web application using Next.js, Tailwind and FastAPI to calculate and visualize drug synergies using interactive data visualizations for cancer treatment research.
- Helped refine drug synergy models by contributing code, analyzing outputs, and proposing algorithmic improvements based on outputs using experimental data from research team.
- Provided technical insights in journal talks and weekly lab meetings to improve research outcomes.
Education
Temple University
M.S. Artificial Intelligence · Beginning Fall 2026
B.S. Computer Science · Graduated May 2026
Studied abroad at Temple University Japan Campus in Fall 2025
Central High School
Philadelphia, PA