Backend / Full-Stack Engineer · AI Applications · AI Native · Full-Stack
System design, architecture, APIs & reliability first · AI application systems · Full-stack · Frontend UI
Focus on the technical essence — architecture, data contracts, reliability, delivery quality —
and treat frameworks, models, and SDKs as a replaceable tool combination.
I focus on the enterprise AI application layer: turn business knowledge into a usable knowledge base, then ship durable workflows — answers with sources, human confirmation before changing data, and an ops trail.
In plain terms: look up company knowledge base → answer with citations → confirm before changing data → leave a reviewable log.
The application backbone stays the same across industries; what changes is the business pack (knowledge base + tools) — helpdesk today, ecommerce or finance tomorrow (e.g. refunds / order updates vs card freeze / trading actions).
Priority order:
- Backend — contracts, data, reliability, security
- AI Native — knowledge base → workflows → standard product shapes
- Full-stack — modern UI when needed with React / Next.js / React Native and other current frontend stacks
Published labs:
- enterprise-ai-lab — swappable business packs on a backend-leaning AI application core (EN/中文 demo)
- fintech-trading-terminal — realtime high-frequency market microstructure UI
- onchain-lab — durable blockchain fundamentals lab
|
API & data contracts · auth / RBAC · idempotency Vehicle: Python · FastAPI · PostgreSQL · Docker · CI |
Component architecture · state · streaming UX Vehicle: React · TypeScript · Next.js · Vitest / Playwright and other current frontend stacks |
Working style — solve the real business problem first; abstract only when a pattern has repeated. Prefer provider-neutral designs so systems survive tool and model churn.
Business: helpdesk demo — answer with sources → confirm before write → ops trail. Same backbone extends to ecommerce (cancel / return / coupon) and finance (freeze card / dispute / place order) via swappable packs. Engineering: FastAPI · PostgreSQL/pgvector · business↔architecture pipeline UI (EN/中文). |
Realtime market data, order book / depth, and durable client patterns — WebSocket lifecycle, batching, throttling, REST fallback. Focus: reconnect / backoff · render coalescing · normalized store · degrade paths |
Provider-neutral lab for durable blockchain fundamentals: keys & signing, reads via ABI / JSON-RPC, transaction lifecycle, on-chain / off-chain sync. Focus: EIP-191/712 · ecrecover · ports & adapters · testnet only / no secrets |
Accounts, transactions, idempotent payment flows, reconciliation, and auditability. Focus: state machines · retry-safe execution · reconcilable ledgers Notes & demos — coming soon |
Architecture notes, tradeoffs, repo layout, API contracts, tests, CI, observability, and ADRs land here as the work is published.
GitHub · campustudio