Dan O.

Key Skills

Software

Don't disclose

Top Subject Matter

Private law / compliant critical data language modeling

Startup product stress-testing and operational intelligence evaluation

Music production metadata accuracy / ethical ML training

Top Data Types

Text

Document

Top Task Types

Fine-tuning

Text Generation

Freelancer Overview

Client (Private Law Firm) — Design & Implementation of local deterministic LM system for critical data (build, training . Brings 23+ years of professional experience across legal operations, contract review, compliance, and structured analysis. Core strengths include Internal, Proprietary Tooling, and Don't disclose. Education includes Native Build - Architecting and Global Operations Director — EarthStream Global /, AI (2015) and Validant / Glandore Systems (, Project Lead (2010). AI-training focus includes data types such as Text and labeling workflows including Fine-tuning, Evaluation, and Rating.

Labeling Experience

theDL.uk — Public spending transparency platform; NLP engine with cited-source, zero-hallucination design (private beta testing)

Text

Built a public spending transparency platform that normalizes and translates government financial data at scale using an NLP engine. Designed the system for zero hallucination by enforcing a cited-source approach where outputs are grounded in verified sources. Conducted private beta testing to validate searchability, verification coverage, and reliability across many government datasets. • NLP normalization/translation over 77 verified UK government sources • Zero-hallucination design via cited-source grounding • Reliability validation through private beta testing • Scaled search and structured access to long-running financial records

2026 - Present

sortit.music — Local-first Mac desktop tool; truth-data focused ML model (ethical ML model training)

TextText Generation

Trained and supported an ethical, truth-data focused local-first ML approach for improving music metadata accuracy. Built a local macOS desktop tool that integrates a truth-focused model to improve metadata correctness. Participated during the licensing/ethical ML training stage prior to broader rollout. • Local-first ML model development for metadata accuracy • Ethical ML training using truth data • Integration into a local desktop workflow (macOS) • Licensing stage preparation for model deployment

2026

Stress-Test — evaluate startup/product/feature ideas before build (assumption testing & risk rating)

Don't discloseText

Developed risk-assessment and stress-testing work for startup product and feature ideas using market and operational assumptions. Conducted structured evaluation to rate risks and identify bias toward failing quickly and cheaply. Focused on determining what would actually happen in the target market before committing to builds. • Assumption extraction from technical/operational/economic hypotheses • Risk rating and pre-build evaluation • Bias-to-fail-cheap-and-fast framework • Pre-deployment feasibility and failure-mode analysis

2026

Client (Private Law Firm) — Design & Implementation of local deterministic LM system for critical data (build, training & tuning)

TextFine-tuning

Designed and built a local deterministic language model system for critical private-law data governed by compliance constraints. Performed build, training, and tuning to adapt model behavior for use cases where frontier transfer was not allowed. Evaluated outputs using a ground-truth and cited-source approach to minimize unsupported generations. • Local deterministic LM system design & tuning • Training and hyperparameter tuning for compliance-bound data • Ground-truth and cited-source evaluation to reduce hallucinations • Red-teaming style testing for robustness and failures

2025

Education

Work History