Job Description & Details
This is a hands-on AI engineering role at DirecTV where you'll be building production-grade GenAI applications that directly touch client-facing streaming platforms. If you enjoy wrangling complex RAG pipelines and integrating LLMs into high-traffic consumer tech like Smart TVs and mobile apps, this gig keeps you right in the middle of modern AI architecture.
What You'll Actually Be Doing
You'll spend your days building and maintaining production-ready GenAI applications using LangChain, LlamaIndex, and Python microservices. A massive part of your bandwidth will go toward implementing high-accuracy Retrieval-Augmented Generation pipelines specifically tailored for sports stats, metadata, and knowledge bases. You'll also deal with the gritty side of LLMs—setting up guardrails, handling model fine-tuning with LoRA/QLoRA on open-source models like Llama, and hooking everything up to DirecTV's client apps via RESTful APIs and WebSockets.
The Core Tech Stack
You need to be dangerous with Python and modern API development using FastAPI or Flask, alongside heavy lifting in LLM orchestration frameworks like LangChain and LlamaIndex. They expect you to know your way around Vector Databases such as Pinecone, Chroma, or Qdrant, and have solid containerization and cloud deployment chops with Docker, Kubernetes, and AWS, GCP, or Azure. If you haven't touched guardrails frameworks like NeMo Guardrails or Ragas for evaluation suites, you're going to hit a steep learning curve here.
Interview Expectations
Expect the hiring manager to grill you on how you handle latency and hallucinations in production RAG systems, likely asking: "Walk me through how you would optimize a RAG pipeline that is currently returning irrelevant chunks and timing out under heavy loads." They want to hear specifics about chunking strategies, reranking, and hybrid search rather than textbook definitions. Another favorite will be testing your grasp on failure modes: expect a deep dive into how you implement circuit breakers and fallback logic when external LLM endpoints throw errors or exceed token limits.
Application Advice
Skip the generic tech buzzwords and make sure your resume explicitly mirrors the tools listed in the JD—specifically highlighting your production experience with FastAPI, LangChain, vector databases, and evaluation frameworks like Ragas or TruLens. If you've built custom agent workflows or fine-tuned open-source models using LoRA, put those front and center because the ATS and the engineering team will be hunting for concrete proof that you've shipped GenAI to production, not just played with notebooks.