Hi, I'm Tiffany 👋

Hi! I'm an AI platform engineer who believes the best way to understand something is to build it. I work end-to-end on production-grade GenAI systems — from polished React/Next.js interfaces to LLM orchestration, multi-agent workflows, and AI gateway infrastructure — and I'm happiest exploring new ideas, rapid prototyping, and turning emerging tech research into things that can ship and scale. ✨

• 🔭 Building enterprise AI gateways, multi-agent platforms, and GenAI workflow tooling
• 🤝 Open-source contributor to OpenClaw and DeepEval
• 🎮 Building side projects — an SDLC multi-agent platform and a gamified DSA learning app
• 💬 Ask me about AI gateways, agentic frameworks, RAG, and LLM evaluation

---

🏢 What I Build at Work

• Enterprise AI Gateway — a unified API letting 100+ internal projects consume 60+ LLMs, embedding, and multimodal models, with auth, auto-routing, caching, load balancing, token-aware rate limiting, and guardrails at low latency.
• Agent Hub — a low-code platform to compose multi-agent teams with configurable routing (auto router, parallel, supervisor), MCP-scoped tools, and reasoning patterns (ReAct, Plan-and-Execute, Reflexion), plus a DeepEval-based evaluation, regression & A/B framework.
• GenAI Workflow Platform — a React Flow + Next.js visual DAG builder for multi-step LLM pipelines (ingestion, RAG, report generation). Automated global screening ops.

---

🚀 Personal Projects

🤖 SDLC Agent

A full-stack multi-agent platform for automating software development workflows — LangGraph · MCP · FastAPI · Next.js.

• Nine-role agent team (Coder, QA, DevOps, Researcher, Planner, Reviewer, PM, Data Analyst, Prompt Optimizer) behind an LLM router
• Orchestration strategies: router_decides, sequential, parallel, supervisor, and auto
• Human-in-the-Loop checkpoints, RAG pipeline, and a LangFlow-style visual workflow builder
• DeepEval-powered evaluation, golden-dataset regression testing, prompt-drift detection, and a PromptOptimizer meta-agent
• Full OpenTelemetry observability with a real-time trace inspector

🎯 DSA Deliberate Play

A gamified Next.js + TypeScript web app that turns DSA practice into deliberate play.

• Interactive algorithm visualizer with real-time animations + synced Python code highlighting
• Gamified challenges across sorting, searching, graphs, linked lists, trees, DP, greedy & bitwise
• XP rewards, progress tracking, achievements, and full dark mode

---

🌱 Open Source

• OpenClaw — merged fix for Discord messages silently dropped during gateway reconnect (#100896), plus Slack, memory, and UI fixes
• DeepEval — fixed token-count & cost population for GPT-5.x in LLM tracing spans, with unit tests

---

🛠️ Tech Stack

AI & Agents: AI Gateway · API & System Design · LLM Fine-Tuning (PyTorch) · Prompt Engineering · RAG · MCP · LangChain · LangGraph · DeepEval · Embeddings Languages: Python · TypeScript · Go · Lua · SQL Frontend: React.js · Next.js · Vue.js · React Native Backend: Microservices · REST APIs · Redis · PostgreSQL · DynamoDB · GCP Pub/Sub Infra: AWS · GCP · Azure · Kubernetes · Docker · Terraform · CI/CD · GitOps

---

📫 Connect