Updated 19h ago

Data Scientist at ABC Legal Services · AI Job Signal

Data Scientist

On-site in Longmont, Colorado. Monday through Friday, business hours, Mountain Time.

About Docketly

Docketly, a sister company to ABC Legal Services, is a fast-growing legal-tech company based in Longmont, Colorado. For the creditors’ rights industry, we make hiring a stand-in attorney easy, fast, and reliable. We pair proprietary software with a nationwide network of attorneys.

Job Overview

You build and run the machine learning and analytics systems behind Docketly’s operations. You build and run production models, from first idea to a live, monitored system on AWS. ABC Legal’s data science team is your sounding board on methods. You build the BI reporting that leadership, operations, and client teams use daily. You support pricing and margin decisions with financial modeling. You work closely with backend engineers on data infrastructure and production issues. This role can grow into Senior Data Scientist as you take full ownership of the production model and the pricing roadmap.

Key Responsibilities

Model development and MLOps

Build and run the full lifecycle of production ML models: framing, feature engineering, training, validation, deployment on AWS SageMaker (Model Registry, real-time endpoints), inference infrastructure (for example, AWS Lambda), and ongoing monitoring.

Build and maintain automated retraining and drift monitoring for deployed models.

Run staged production rollouts with holdout comparisons to prove model impact before full release. Review the test design with the team before launch.

Use AI coding tools to speed up analysis, modeling, and MLOps work.

Analytics, pricing, and reporting

Build and maintain the BI layer leadership relies on for exec reporting on revenue, margin, and volume, from dashboard design through data accuracy.

Run pricing and margin analysis for client accounts, including ad hoc investigations for named strategic accounts. Model the financial impact of proposed pricing changes before they go live.

Build and maintain rules-based pricing and case-difficulty logic by court and county, alongside statistical and ML approaches.

Build and tune large scoring and prioritization frameworks in Metabase, such as weighting, percentile scenarios, and drop-rate analysis, that drive daily operational decisions such as Hearing priority and assignment.

Turn ambiguous business questions into clear, well-scoped analyses.

Cross-functional partnership and documentation

Partner with engineering on data pipelines (AWS Glue, Lambda, EventBridge). Investigate and help resolve production database performance issues, such as ORM query patterns and indexing, alongside backend engineering.

Write down your methods and decisions in Confluence or an equivalent shared space, as you go, so other teams can find your reasoning and build on it without asking you directly.

Systems & Skills

Required

Bachelor’s degree in a related field, or equivalent practical experience, and 3+ years of experience with applied statistics or machine learning.

Applied statistics: probability, hypothesis testing, and model evaluation across classification and regression/quantile models (for example, precision, recall, calibration, quantile coverage, and drift).

3+ years of experience with SQL and a relational database in a professional capacity. Docketly’s production database runs on MySQL/Aurora, about 400 tables.

Python for data analysis and modeling: pandas, scikit-learn, and a gradient-boosting framework (XGBoost, LightGBM, or CatBoost). Our production model uses CatBoost-based quantile regression.

Experience training and deploying at least one model to production, or on a cloud ML platform such as AWS SageMaker.

Metabase or comparable BI tooling.

Clear written and verbal communication with non-technical stakeholders.

Preferred

Staged production rollouts (canary or phased release) with holdout comparisons to prove model impact before full release.

AWS Glue, S3, Lambda, EventBridge, CloudFormation.

Comfortable using AI coding tools such as Claude Code or GitHub Copilot to move faster on analysis and modeling work.

Experience diagnosing production database performance issues, such as query plans, indexing, and ORM-generated query patterns, in a large, high-table-count schema.