Accurate HCC coding is critical for patient risk adjustment and can have a material impact on the revenue integrity of healthcare systems and payers. Multiple studies have shown that one third to one half of patients can have evidence of prior conditions, complications, or severity in their clinical notes that are not reflected in claims or EHR problem lists. This talk presents an automated solution which automates the discovery of missed clinical codes from free‑text patient notes, prioritizes them based on their potential impact to each patient’s risk adjustment score, and provides a human‑in‑the‑loop validation tool for manual review and audit. Its key benefit is enabling organizations bring AI‑powered HCC coding in‑house – empowering existing teams with greater control, scalability, and cost efficiency. The solution architecture is based on a combination of state‑of‑the‑art, healthcare‑specific language models by John Snow Labs. A technical deep dive will share lessons learned from combining LLM and NLP models to deliver the accuracy, scalability, and explainability that were required by the enterprise who have deployed it.