Keith JungIntegrations Lead
Jiwon (Keith) Jung · Integrations Lead, legal tech
Clients to data. Tools to tools. Automation to automation. Wherever an integration is needed, that’s where I come in, from the first scoping call to the system that keeps running after launch.
01 · The gap
Clients keep their data in one place. Teams work in another. Automations fire into a third. Most of the value sits where those places meet, and that’s the part I own.
Law firms stay in the case system they already use, so the integration has to come to them. I scope requirements directly with each firm, ship the API layer, and stay with it after launch.
Two platforms rarely agree on what a “matter” is. I reconcile the data models, own the mapping, and watch delivery so a broken sync surfaces on a dashboard instead of on a client call.
Intake that routes itself. Dashboards that refresh themselves. Alerts that reach the Account Executive before the client picks up the phone. I chain the pieces so the whole workflow runs end to end.
03 · In production
What runs today, built from the client’s problem to a shipped system, and monitored after launch.
Salesforce Litify, Filevine and Monday.com each have their own idea of a case. I scope with each firm, build the API layer, reconcile the models, and monitor every delivery.
Leads were stuck in a call queue. With Marketing, I shipped a self-service intake form and ran the test behind it, with media spend per matter held equal.
Elasticsearch checks on every integration, SQL-backed executive dashboards, and a live status view so Account Executives answer client questions on the spot. Client reporting went from days to same-day. Break something and watch what happens:
On mass-tort dockets (Talc, AFFF, Camp Lejeune), the evidence sat in unstructured case documents and every firm documented differently. I built Python NLP models that read the documents and turn them into cashflow and collateral health signals. The underwriting team uses them in live decisions.
04 · Under the hood
A mathematics and computer science foundation, machine learning research, and eighteen months leading a squad. The spec sheet:
Owns client integrations from scoping to launch and monitoring. Builds the internal tools teams use to decide: dashboards, status views, risk signals.
Consolidated multi-source patient data without dropping conflicting rows, then fine-tuned a TensorFlow model by transfer learning. First place at the Korean Society of Laboratory Medicine Summer Symposium.
Trained models on cell averages, paired with finite volume methods, to cut compute cost and step past CFL limits.
Squad Leader for eighteen months. Assigned roles by individual strength and held one standard. Team awarded “Top Team”.
REST API integrations and data cleaning on Google Cloud Platform, contributing to a 32% sales increase.
Converted Excel and JSON sources to SQL through an ETL process for a data-mining project that drove a 28% sales increase.
05 · Connect
A client and their data. A tool and the one next to it. An automation that stops where another should start. Tell me about it.