.webp)
Resumes
August 13, 2026
Resume Example
Real-world NLP engineer resume examples across legal, clinical, and search roles, with the F1 scores, latency, cost, and business outcomes that machine learning teams look for.
Linh Truong
(617) 555-0147 ◇ Boston, MA
linh.truong@example.com ◇ linkedin.com/in/linh-truong ◇ linhtruong.com
Objective
NLP engineer with 5 years building text classification, extraction, and search models, currently at a legal technology company whose contract review product processes 400,000 contracts a year. Built the clause extraction model that reached 0.91 F1 across 38 clause types, cut review time per contract from 50 minutes to 12, and lowered inference cost 58% by distilling a large model into a smaller one. Cares about clean evaluation sets and models that lawyers actually trust.
Education
M.S. in Computer Science, Charles River Crown University 2019 – 2021
Focus on Natural Language Processing. Coursework in Statistical NLP, Deep Learning, Information Retrieval, and Machine Learning Systems.
B.S. in Mathematics and Linguistics, Bay State Line University 2015 – 2019
Honours thesis on dependency parsing for low-resource languages.
Skills
NLP Methods
Named entity recognition, span extraction, text classification, semantic search, sentence embeddings, transformer fine-tuning, knowledge distillation, active learning
Languages and Libraries
Python, PyTorch, Hugging Face Transformers, spaCy, scikit-learn, sentence-transformers, pandas, FastAPI
Evaluation and MLOps
Annotation guidelines, inter-annotator agreement, precision and recall analysis, error analysis, experiment tracking, model serving, drift monitoring
Tools & Platforms
AWS SageMaker, Docker, Weights and Biases, Label Studio, Elasticsearch, ONNX Runtime, PostgreSQL, GitHub Actions
Experience
NLP Engineer 02/2023 – Present
Beacon Hill Line Legal Tech Boston, MA
- Built a clause extraction model reaching 0.91 F1 across 38 clause types for a product reviewing 400,000 contracts a year.
- Cut average review time per contract from 50 minutes to 12 minutes, measured across 14 enterprise customers.
- Lowered inference cost 58% by distilling a large transformer into a smaller model served with ONNX Runtime.
Machine Learning Engineer, NLP 07/2021 – 01/2023
Harbor Crown Software Cambridge, MA
- Built a support ticket classifier routing 25,000 tickets a week into 45 queues at 89% accuracy, up from 71% with rules.
- Shipped semantic search over 12,000 help articles that raised self-service resolution from 22% to 34%.
- Set up an active learning loop in Label Studio that cut labelling needed for new ticket categories by about 60%.
Graduate Research Assistant, NLP Lab 09/2019 – 05/2021
Charles River Crown University Boston, MA
- Co-authored 2 workshop papers on cross-lingual entity recognition for 6 low-resource languages.
- Built an annotation pipeline and guidelines used by 8 annotators to label 30,000 sentences at 0.84 agreement.
- Taught 2 lab sections of an NLP course for about 70 students, running weekly coding sessions in PyTorch.
Projects
Clause Extraction Model. Wrote annotation guidelines with 3 contract lawyers, labelled 9,000 contracts across 38 clause types, and fine-tuned a legal-domain transformer for span extraction with per-clause thresholds, which reached 0.91 F1 and cut contract review time from 50 minutes to 12 minutes.
Model Distillation for Serving. Distilled the production extraction model into a model 4 times smaller, exported to ONNX and quantised, losing only 0.01 F1, which cut inference cost 58% and p95 latency from 1.9 seconds to 0.6 seconds per contract page.
Evaluation Harness. Built a regression test set of 1,500 hand-checked contracts with per-clause scoring and an error analysis report run on every model change in GitHub Actions, which caught 3 releases that would have dropped recall on rare clauses.
Extra-Curricular Activities
- Reviewer for 2 NLP workshops on low-resource languages, reviewing about 6 papers a year.
- Contributor to an open source Vietnamese NLP toolkit, with 9 merged pull requests on tokenisation.
- Speaker at a Boston machine learning meetup on evaluating extraction models, 2 talks since 2023.
Leadership
- Lead the annotation programme for the NLP team, managing guidelines for 5 contract annotators.
- Mentor 2 junior machine learning engineers on evaluation design and error analysis each week.
- Run a monthly paper reading group for 12 engineers across the machine learning and search teams.
.webp)
%20Which%20Should%20You%20Use.webp)
