Cindy L.

Production AI Engineer | LLM, RAG & Backend Integration I help founders and product teams build and improve LLM applications, RAG systems, and AI-enabled backend workflows—with a focus on reliable data, maintainable code, and safe integration into existing software. As a Top Rated Upwork freelancer, I bring hands-on experience supporting an AI-driven financial research platform through backend integration, technical review, production debugging, and release validation. 📣 Client Feedback — Python & Hospital Scheduling Project “Cindy completed work on a complex Python project with advanced business logic. I found her to be very thorough and pleasant to work with.” ✅ Commercial Experience — AI Financial Research Platform Ongoing, long-term Upwork engagement I contribute to the development and production hardening of an existing AI-driven research platform, working across LLM workflows, external financial-data integrations, report generation, APIs, PostgreSQL-backed persistence, and access controls. My contribution combines backend integration, implementation review, troubleshooting, and validation across application and data boundaries. Key responsibilities include: • Reviewing and validating AI/backend changes against requirements, tests, and actual system behavior. • Checking LLM-generated reports against authoritative source data, including data provenance, missing evidence, and consistency between stored records and displayed outputs. • Investigating data-ingestion failures, API issues, database connectivity problems, and differences between deployed environments. • Validating authentication and entitlement behavior, database changes, and API compatibility. • Reviewing pull requests, running regression checks, and identifying implementation risks before release. • Supporting pre-production deployments and bounded data backfills, with execution logs and validation evidence. • Coordinating with frontend, backend, data, and product contributors to clarify dependencies and preserve existing behavior. This work has given me practical experience with the challenges of maintaining an evolving AI product: tracing failures across services, distinguishing data problems from presentation issues, and validating changes within an established codebase. ✅ Selected Portfolio & Project Work Financial QA Assistant — RAG Extension (Open-Source Base) Extended the open-source rag-chatbot (Apache-2.0) into a financial QA assistant, adding SEC filing ingestion with just-in-time 8-K/10-Q indexing, financial-data APIs (SEC XBRL, Alpha Vantage, Yahoo Finance, NewsAPI), Tools → RAG → Demo → LLM routing, and a Claude fallback. Built with Python, Chroma, llama.cpp and Streamlit; I also extended inherited chat-history and retrieval logic. Healthcare AI Platform — Portfolio Project in Development Developing a healthcare workflow platform that combines structured intake processing, retrieval-supported AI outputs, safety checks, and PostgreSQL persistence. The current work focuses on backend workflows and data handling. Further development is directed toward evaluation, traceability, and demonstrating a complete application workflow. This is an independent portfolio project in development. Timefold-Based Hospital Scheduler — Backend & Constraint Optimization Built a hospital shift-scheduling backend using Python, FastAPI, and the Timefold constraint solver. Implemented business rules covering availability, skills, shift groups, and contract constraints, together with JSON APIs, Dockerized deployment, and cURL/Pytest validation. This project demonstrates my ability to translate complex operational requirements into testable backend behavior. Data Science & Predictive Modeling Built analytical and machine-learning pipelines using Pandas, NumPy, Scikit-learn, and SQL, covering data cleaning, feature engineering, classification, regression, and model evaluation. This foundation supports AI applications that require structured data processing and conventional ML alongside LLM workflows. 🛠 What I Can Help You With • LLM & RAG Applications Document processing, embeddings, vector search, source-grounded answers, structured outputs, and LLM integration. • AI Backend & API Integration Python/FastAPI services, PostgreSQL/SQL workflows, REST APIs, and connections between AI capabilities and existing business systems. • Existing AI System Improvement Reviewing implementation quality, investigating failures, integrating fixes, and validating changes within an established application. • AI Reliability & Validation Source-data checks, output validation, regression testing, failure analysis, and review of logging, metrics, and traceability. • Data Processing & Business Logic Data pipelines, predictive modeling, and constraint-based optimization for operational workflows. 🤝 How I Work I start by clarifying the workflow, existing system, available data, and acceptance criteria. I break implementation into reviewable deliverables