W042021–25AI enablementClinical trials, pharma9 trials
Outcome
80%
faster response time from the RAG document intelligence system, built across structured and unstructured clinical data, HIPAA-compliant throughout.
200,000dollars in annual operational inefficiency eliminated by a full JIRA instance restructuring
9decentralized clinical trials led end to end, provisioned and BYOD devices, globally
3Fortune 500 pharmaceutical clients run concurrently

Nine trials, 18 countries, one document system that got 80% faster

01 / Context

End-to-end delivery for 9+ decentralized clinical trials — custom mobile application development for both provisioned and BYOD devices, globally.

One reference program alone, for the largest of the three pharmaceutical clients, ran 4 clinical trials as a single $8M fixed-budget program, 18 countries, 400+ provisioned iOS devices, teams spanning the US, Croatia, Argentina, and Poland.

02 / Problem

Clinical operations teams were searching structured and unstructured data by hand, in a space where getting the answer wrong isn’t an option.

Each trial’s mobile applications needed to be localized and meet trial-specific protocols and compliance requirements — for Fortune 500 pharma clients, in a regulated environment, across multiple time zones.

DOCUMENT INTELLIGENCE PIPELINE — DIAGRAM
FIG 01 — Structured and unstructured clinical data, one retrieval layer.

03 / Approach

Design for the regulated environment first, and treat the platform underneath it as part of the same program.

01Design and implement a RAG-based document intelligence system over both structured and unstructured clinical data.
02Keep HIPAA-compliant data handling non-negotiable through every stage of the build.
03In parallel, restructure the JIRA instance completely — the operational backbone the whole program ran on.

04 / What I did

Owned design, rollout, and adoption of LLM-powered assistants supporting clinical operations and customer workflows across multiple trial programs.

Contributed directly to model optimization and fine-tuning — structuring training data, refining prompts, and guiding evaluation criteria aligned to clinical and operational use cases. Orchestrated the complete JIRA restructuring that eliminated $200K+ in annual inefficiency.

Role: Sr. Agile Project Manager · Clinical-trial technology company · Nov 2021–Feb 2025 · 18 countries, 400+ devices

05 / Outcome

The RAG system cut response time by 80% while holding HIPAA-compliant data handling throughout. The JIRA restructuring paid for itself in visibility as much as dollars — $200K+ in eliminated inefficiency plus improved cross-team visibility across every concurrent trial.

RESPONSE TIME, BEFORE/AFTER — DATA VISUAL
FIG 02 — Response time, 80 percent faster, HIPAA-compliant throughout.

06 / What I would do differently

I’d have built a per-country localization and compliance checklist after the first trial, not relearned the same constraints fresh on every one of the next eight — the pattern repeated more than the plan accounted for.