Portfolio / Brooklyn, NY, United States

Stephen Liu

Senior AI Engineer

Senior AI Engineer with 8+ years of experience building production-scale AI systems focused on large language models (LLMs), retrieval-augmented generation (RAG), semantic search, recommendation systems, and agentic AI workflows. Strong expertise in designing end-to-end AI systems using Python-based backend services, vector databases, and modern LLM frameworks.

Experienced in building scalable AI-driven products for search, discovery, and personalization in production environments.

+1 (202) 922-5639 stephenl583721@gmail.comBrooklyn, NY, United States

View Projects Get In Touch

01 / Experience

Professional History

Eight years building production-scale AI systems for search, discovery, personalization, and agentic workflows.

May 2024 — May 2026

Mindtrip, Inc

Senior AI Engineer

Python
Node.js
LangGraph
LangChain
RAG
FastAPI
Elasticsearch
pgvector
Redis
AWS

Built and scaled an AI-powered travel booking platform enabling end-to-end itinerary generation from natural language queries using LLMs and multi-agent systems
Designed agentic workflows using LangGraph for multi-step planning, reasoning, and validation across travel itinerary generation
Developed backend orchestration services in Python and Node.js to manage LLM pipelines, retrieval systems, and request routing
Implemented RAG-based retrieval and hybrid search systems combining structured travel data with semantic embeddings for improved result grounding
Built multi-stage ranking and reasoning pipelines to optimize travel results based on user constraints such as budget, time, and location
Integrated LLM-driven intent detection, structured output generation, and evaluation loops to improve response accuracy and consistency

Feb 2020 — Apr 2024

Constructor

AI Engineer

Python
FastAPI
Node.js
LangChain
LangGraph
Elasticsearch
FAISS
Redis
AWS

Built and scaled enterprise-grade AI search and recommendation systems for large-scale e-commerce product discovery and semantic search
Designed and developed backend services for retrieval, ranking, and orchestration layers supporting hybrid search (BM25 and vector-based retrieval)
Implemented embedding-based search pipelines and transformer-based ranking models to improve relevance and query understanding
Worked on multi-stage ranking systems combining traditional learning-to-rank methods with modern neural and LLM-based approaches
Built APIs and backend services to handle high-throughput search traffic with low-latency requirements
Integrated caching and optimization strategies to improve performance and scalability of search systems
Contributed across the full AI stack including retrieval systems, backend services, and production deployment support

Sept 2017 — Dec 2019

TechNova Systems

Full Stack Engineer

React
Node.js
Python
PostgreSQL
MongoDB
AWS
Jenkins
GitHub Actions

Built and maintained full-stack web applications supporting search and recommendation features
Developed backend services using Node.js and Python for API development and data processing
Designed and implemented responsive React-based frontend applications
Built scalable REST APIs supporting web application functionality and real-time data flow
Designed and optimized relational and NoSQL database structures (PostgreSQL, MongoDB)
Implemented authentication systems and session management across applications
Deployed applications on AWS (EC2, S3) and maintained CI/CD pipelines using Jenkins and GitHub Actions

02 / Portfolio

Previous Projects

Production AI travel planning and e-commerce search systems with RAG, LangGraph, and hybrid retrieval.

AI / Travel / May 2024 — May 2026

Agentic AI Travel Planning System

Mindtrip, Inc

Built a production-grade AI travel booking platform powered by LLMs and multi-agent orchestration, enabling personalized end-to-end travel itinerary generation from natural language queries
Designed a hybrid retrieval architecture combining Elasticsearch (BM25) and PostgreSQL with pgvector to support both structured and semantic search across flights, hotels, and activities
Implemented a multi-stage ranking pipeline using Hugging Face cross-encoder rerankers and GPT-4o-based contextual ranking to optimize results based on user intent, budget, geography, availability, and temporal factors
Designed and implemented a multi-agent orchestration system (planner, retriever, validator, synthesizer) using LangChain for LLM workflow orchestration and LangGraph for stateful, graph-based execution of complex travel planning workflows
Built a scalable backend orchestration layer using FastAPI and Node.js (Express) for request routing, authentication (JWT), Redis caching, and coordination across retrieval, ranking, and LLM generation services
Integrated external travel and mapping APIs including Amadeus, Skyscanner, Expedia, Booking.com, and Google Places with rate-limited and fault-tolerant API handling
Developed robust ETL pipelines using Python and Apache Airflow for ingestion, normalization, and embedding generation using OpenAI embedding models
Implemented production-grade system capabilities including Redis caching, API rate limiting, observability using OpenTelemetry and Prometheus, distributed tracing, and fault-tolerant orchestration

LangGraphLangChainRAGFastAPINode.jsElasticsearchpgvectorRedisAWSAirflow

Live Demo

AI / E-Commerce / Feb 2020 — Apr 2024

AI Commerce Search & Recommendation System

Constructor

Built a production-grade AI-powered e-commerce search and product discovery system that evolved from classical ML-based search into modern LLM-augmented search and ranking systems, improving relevance, personalization, and conversion for large-scale product catalogs
Designed a hybrid search architecture combining Elasticsearch (BM25-based lexical search) and vector-based semantic search to support keyword-driven and natural language intent-based queries
Implemented a multi-stage ranking pipeline using classical learning-to-rank approaches, cross-encoder transformer models, and LLM-based reasoning with LangChain and LangGraph for stateful multi-step ranking workflows
Developed a real-time query orchestration service using FastAPI and Node.js to manage search requests, route queries across retrieval and recommendation systems, and integrate ML ranking models with LLM reasoning layers
Built a clickstream-driven personalization system evolving from rule-based ranking to behavior-driven dynamic ranking using clicks, views, and purchase signals for session-aware personalization
Designed scalable data and indexing pipelines using AWS S3, PostgreSQL, and ETL workflows to process product catalogs and maintain dual indexing in Elasticsearch and vector databases
Deployed the system on AWS using ECS, Lambda, and S3 with Docker-based microservices architecture and CI/CD pipelines optimized for low-latency, high-throughput search traffic

PythonRAGLangChainLangGraphFastAPINode.jsElasticsearchFAISSRedisAWS

Live Demo

01 / 00

03 / Skills

Technical Competencies

LLM systems, hybrid search, vector databases, and production AI infrastructure across travel and e-commerce.

AI / LLM Systems

Large Language Models (LLMs)
Retrieval-Augmented Generation (RAG)
Prompt Engineering
Embeddings
Semantic Search
Agentic AI Systems
LangChain
LangGraph
Evaluation Pipelines

Search & Retrieval Systems

Hybrid Search (BM25 + Vector Search)
FAISS
Pinecone
Weaviate
Vector Databases
Ranking Systems
Information Retrieval

Backend Engineering

Python
FastAPI
Node.js
REST APIs
Microservices
Asynchronous Processing
System Orchestration

Frontend

React
Angular
Vue.js

Cloud & DevOps

AWS (EC2, S3, Lambda)
Docker
CI/CD
GitHub Actions
Jenkins

Databases

PostgreSQL
MySQL
MongoDB
Vector Databases

04 / Education

Academic Background

Bachelor's degree in computer science with foundations in software engineering and systems.

Sept 2014 — May 2017

New Rochelle, New York

University

Iona University

Bachelor's Degree in Computer Science

05 / Contact

Get In Touch

Open to senior AI engineering roles, contract work, and LLM platform collaborations.

Phone

+1 (202) 922-5639

stephenl583721@gmail.com

Location

Brooklyn, NY, United States