Four years of legal practice.
Six systems built along the way.

The Challenge

A complex real estate restructuring required consolidating highly fractured, multi-generational fractional property interests. A multi-phase swap-and-sale strategy was developed to mitigate tax exposure, leveraging intrafamily exemptions and tax-deferred exchanges — which meant generating over 400 unique, interconnected deeds. Variables had to update correctly across four consecutive phases, including shifting fractional share calculations and rotating parties. Drafting by hand was slow and carried a high risk of data mistakes.

The Solution

Designed and built a custom automated document assembly system driven by a relational database backend and a Python-based document generation script, using self-validating data structures to map relationships between parties, parcels, and transaction phases.

Workflow Architecture

• Centralized Relational Database — Extracted raw property, party, and fractional share data into a single source of truth linking properties to parties and phases. Embedded logic automatically calculated shifting fractional interests and swapped Grantor/Grantee roles as phases progressed.
• Automation Controller — A Python script acted as middleware, querying the database for phase-specific data and routing it to the correct legal templates.
• Batch Document Assembly — The script dynamically drove Microsoft Word via VBA, injecting structured data into parametric .docx templates — autonomously generating and filing 400+ signature-ready deeds and supporting documents in minutes.

The Outcome

Reduced the transaction cycle from weeks of manual drafting to automated generation in minutes. Because the underlying data was relational and self-validating, the entire document corpus could be mass-regenerated whenever deal variables or phase structures changed — eliminating data drift and manual error.

The Challenge

A team of 55 remote attorneys drafted title opinion letters for properties nationwide, relying on scattered, multi-tabbed spreadsheets to track compliance requirements across 5,000+ local jurisdictions. The underlying data was unstructured and maintained by dozens of individual attorneys — data sparsity, spelling errors, and overlapping jurisdiction names (dozens of distinct "Jefferson County" entities nationwide) prevented straightforward translation into a unified database. Spreadsheets that were often locked or took minutes to load created operational bottlenecks and repetitive inquiries in chat channels.

The Solution

Architected a comprehensive data extraction and normalization pipeline to transform disparate, messy datasets into a centralized, relational knowledge-management database.

Workflow Architecture

• Entity Resolution & UID Engineering — Resolved naming conflicts and misspellings by engineering a universal indexing system using Federal Information Processing Standard (FIPS) codes as unique primary keys for every U.S. county.
• Automated Data Extraction & Mapping — Designed automated workflows to parse the varied structures of legacy Excel files, matching FIPS codes to localized data points regardless of the original document's formatting.
• Normalization & Sparsity Management — Mapped, linked, and deduplicated massively redundant and sparse fields (averaging just 10 of 40 potential columns populated per jurisdiction), normalizing unstructured text into a clean relational schema.
• Secure User Deployment — Built a form-based front end with strict access controls and input constraints, democratizing the data for less technical attorneys while fully insulating the backend from unintended edits.

The Outcome

Converted fragmented, unstructured spreadsheets into a scalable, single-source-of-truth relational database. Eliminated spreadsheet lockouts entirely, let 55 remote attorneys obtain and update information in seconds, and created a scalable architecture for future data integration.

The Challenge

Managing a growing trademark portfolio required tracking deadlines across state and federal jurisdictions. Calculating statutory deadlines — USPTO Section 8 & 15 six-year filings, Section 8 & 9 ten-year renewals — was done manually, on paper, for each registration date. The manual process made tracking difficult, created busywork, and increased the risk of missed deadlines that could lead to accidental abandonment of client marks.

The Solution

Built a centralized trademark tracking database integrating client records, office actions, jurisdictional data, registration statuses, and direct USPTO links. Entering a serial number automatically pulls live data from the USPTO API — application status, publication dates, registration dates — and generates a dynamic link to the USPTO TSDR case page. At the core is an automated workflow that reads registration dates, determines precise color-coded maintenance windows, and sorts them into a priority queue for the legal team.

Workflow Architecture

• Data Ingestion Pipeline — Entering a trademark serial number triggers an automated query against the USPTO TSDR API to pull live metadata.
• Centralized Relational Database — Integrates client records, jurisdictional rules, and live USPTO metadata, generating dynamic links directly to the USPTO TSDR case page.
• Automated Alerting & Queueing — A script reads registration dates to calculate precise, color-coded statutory maintenance windows, feeding them into a priority "Upcoming Dates" queue.

The Outcome

Replaced the legacy analog system, modernizing portfolio management and improving the reliability of the firm's IP tracking. The "Upcoming Dates" queue gives attorneys and paralegals an at-a-glance view of upcoming filing windows, reducing the risk of missed deadlines.

The Challenge

Corporate Transparency Act (CTA) implementation required the firm to establish new, secure methods for collecting and managing highly sensitive Beneficial Ownership Information (BOI).

The Solution

Served on the firm's CTA committee and designed the internal intake systems and data-handling policies for beneficial ownership information. The transactional/corporate practice group rarely managed data this sensitive and had no established workflow for it. To bridge the gap, I adapted a dedicated secure software platform already used by the litigation department, learned it myself, and trained our corporate attorneys and paralegals on how to use it.

The Outcome

Standardized the transactional practice group's data collection processes for sensitive personal information.

The Challenge

A team of attorneys was asked to deliver feedback on due diligence documents, but the existing reporting process was entirely manual — drafting individual emails, looking up recipients per file, and typing out file and error details by hand. The friction created bottlenecks during high-volume transaction cycles, leading to low feedback rates. What feedback was submitted stayed trapped in unstructured email chains, preventing management from aggregating data or tracking error trends.

The Solution

Designed and coded an interactive Adobe Acrobat PDF feedback form with embedded custom JavaScript — checkboxes for common document errors plus a free-text field for notes. A button inside the form automatically generated a Microsoft Outlook email, populated with the correct recipients and a subject line built from the data entered in the PDF. The attorney would click the button, send the email, and get back to work. This structured what managers received, making it easy to review and map toward a future web-based reporting dashboard.

The Outcome

Significantly reduced operational friction and increased feedback submission rates. Moving from unstructured emails to a self-populating form demonstrated how embedding automation directly into a team's existing workspace drives adoption for administrative work that sits outside people's core job duties.

The Challenge

When an internal generative AI chatbot was introduced company-wide, adoption in the corporate counsel department was hindered by skepticism and mistrust — attorneys perceived the tool as a "black box."

The Solution

Collaborated with department management to design and deliver a series of AI education sessions for a team of 55 attorneys. We used practical analogies and visual aids to explain abstract concepts and provided a safe, no-judgment space for Q&A. The curriculum covered:

• The core differences between traditional automation and modern large language models.
• The technical mechanics of how chat interfaces process data, predict language, and surface information.
• The limitations of AI systems, including data-currency issues and accuracy risks.

The Outcome

Audience feedback indicated a firmer grasp on how generative AI actually works and a greater — if appropriately cautious — curiosity to explore its uses, helping overcome initial skepticism and drive interest in adoption.