Looking for a job

Amy Zeller

@amyzeller

Senior Full-Stack AI Engineer at DellVerx who built a production RAG assistant that cut support ticket resolution time by 35%.

United States

Message

At DellVerx, I built a production RAG assistant over 15K+ knowledge base articles, reducing average support ticket resolution time by 35%. I also improved answer relevance from 71% to 92% and cut inference costs by 31% through retrieval tuning, evaluation, caching, and model routing.

Previously, I fine-tuned a Llama model at BlackByteSolutions and built a document ingestion pipeline powering semantic search. At 10Pearls, I developed Python APIs and prototyped a chatbot using the OpenAI API; my projects include a multi-agent research assistant and an enterprise document Q&A platform.

Experience

Work history, roles, and key accomplishments

Current

Senior Full-Stack Engineer (Python / GenAI)

Current

DellVerx

Jan 2023 - Present (3 years 9 months)

Built a production RAG assistant with Python, FastAPI, LangChain, OpenAI, and Pinecone over 15K+ knowledge base articles, reducing average support ticket resolution time by 35%. Improved answer relevance from 71% to 92% by tuning chunking, adding hybrid search and re-ranking, and automating evals with RAGAS and LangSmith.

Python fastAPI Lang Chain OpenAI Pinecone RAG Hybrid Search Reranking Ragas Langsmith LangGraph Tool calling Semantic Caching (GPTCache)Prompt Compression Model Routing Mentorship

Full-Stack Developer

BlackByteSolutions

Aug 2021 - Dec 2022 (1 year 4 months)

Fine-tuned an open-source Llama model with LoRA/PEFT for domain text classification, matching GPT-4 accuracy at roughly one-tenth of the per-call cost. Built a Python ingestion pipeline (Celery, embeddings, pgvector) processing 50K+ documents to power semantic search across the core product dashboard.

Python LoRAs PEFT Celery Pgvector fastAPI SSE React PostgreSQL MongoDB Query Optimization Semantic Search

Software Engineer

10Pearls

Jun 2020 - Jul 2021 (1 year 1 month)

Developed and maintained RESTful APIs in Python (Django, Flask) and Node.js for a multi-tenant SaaS platform, supporting a 3x increase in active accounts without degradation. Prototyped the company's first chatbot using the OpenAI API and a retrieval layer over FAQ content, validating demand for a dedicated GenAI roadmap.