Prompt Engineer Resume Examples

Noor Al-Amin

(650) 555-0161 ◇ Mountain View, CA

noor.al-amin@example.com ◇ linkedin.com/in/noor-al-amin ◇ nooralamin.com

Objective

Prompt engineer with 4 years designing, evaluating, and maintaining prompts for production language model features, currently the lead prompt engineer for a customer support product handling 2.4 million conversations a month across 14 languages. Raised automated resolution rate from 31% to 58% through prompt and retrieval redesign, cut hallucinated policy answers 87% measured on a 3,000-case evaluation set, and built the prompt evaluation harness now used by 6 product teams. Maintains a library of 140 versioned prompts with regression tests on every change. Treats prompts as code: versioned, tested, measured, and never trusted on a good demo alone.

Education

M.S. in Computational Linguistics, Peninsula Crown University 2019 – 2021

Coursework in Natural Language Processing, Statistical Language Models, Semantics and Pragmatics, Machine Learning, and Evaluation Methods.

B.A. in Linguistics and Computer Science, Bay Line University 2015 – 2019

Graduated with honours. Coursework in Syntax, Discourse Analysis, Algorithms, Databases, and Human-Computer Interaction. Fluent in Arabic and English with working French.

Skills

Prompt Engineering

System and task prompt design, few-shot and structured output prompting, retrieval-augmented generation prompt design, tool and function calling prompts, multilingual prompt adaptation, safety and refusal behaviour design, prompt versioning and regression testing

Evaluation and Analysis

Evaluation set construction and labelling guidelines, automated and human evaluation design, model-as-judge calibration, error analysis and taxonomy building, A/B testing of prompt variants, cost and latency analysis, hallucination measurement

Engineering and Collaboration

Python for evaluation pipelines and data analysis, working with model APIs, collaboration with product managers, engineers, and support operations, documentation and prompt style guides, training teams on prompting practice

Tools & Platforms

Python with pandas, Claude and other model APIs, LangSmith and Braintrust, Weights and Biases, Jupyter, Git, SQL, Label Studio, Jira, Notion

Experience

Lead Prompt Engineer, Customer Support AI 05/2023 – Present

Peninsula Crown Software Mountain View, CA

Lead prompt engineering for a customer support product handling 2.4 million conversations a month across 14 languages for 900 business customers, on an applied AI team of 11.
Raised automated resolution rate from 31% to 58% through a system prompt redesign, a retrieval prompt with citation requirements, and 40 prompt iterations validated on a 3,000-case evaluation set.
Cut hallucinated policy answers 87% and escalation misroutes 45%, and built the prompt evaluation harness with 12 automated metrics now used by 6 product teams across 140 versioned prompts.

Prompt Engineer and Conversation Designer 07/2021 – 04/2023

Bay Line Conversational AI Palo Alto, CA

Designed prompts and conversation flows for 8 enterprise assistant deployments over 22 months, reaching about 600,000 users.
Built a multilingual prompt adaptation process for 9 languages that held quality within 5 points of English on a 500-case evaluation.
Cut average prompt token cost 38% through prompt compression and caching strategies without measurable quality loss.

Natural Language Processing Research Assistant 09/2019 – 06/2021

Peninsula Crown University Stanford, CA

Built evaluation datasets of 12,000 annotated examples for 2 language model research projects over 2 years.
Co-authored 2 conference papers on evaluating generated text with 1 selected for oral presentation.
Designed an annotation guideline that raised inter-annotator agreement from 0.61 to 0.82 across 6 annotators.

Projects

Support Assistant Prompt Redesign. Rebuilt the support assistant prompt system with a layered structure of policy, persona, and task prompts, a retrieval prompt requiring cited sources for every policy claim, structured output for handoff decisions, and 40 iterations tested on a 3,000-case evaluation set with human review of 500, which raised automated resolution from 31% to 58%, cut hallucinated policy answers 87%, and saved an estimated $6M a year in support cost.

Prompt Evaluation Harness. Built an evaluation harness with 12 automated metrics including accuracy against gold answers, citation validity, refusal correctness, and format compliance, a calibrated model-as-judge checked against 1,200 human labels, and regression runs on every prompt change, adopted by 6 product teams, which cut prompt-related production regressions to 0 across 18 months and cut evaluation turnaround from 3 days to 40 minutes.

Multilingual Prompt Adaptation. Designed a process for adapting prompts to 14 languages with native-speaker review, language-specific few-shot examples, and a per-language evaluation set of 200 cases, which held quality within 4 points of English across all 14 languages and enabled launches in 6 new markets in 9 months.

Extra-Curricular Activities

Speaker at 3 applied AI meetups on prompt evaluation and multilingual prompting.
Maintain an open-source prompt evaluation toolkit with about 1,500 stars on GitHub.
Volunteer Arabic language tutor for a refugee support nonprofit, about 3 hours a week.

Leadership

Lead prompt engineering for an applied AI team of 11, owning the prompt library, review process, and evaluation standard.
Trained 40 engineers and product managers across 6 teams on prompt design and evaluation through 8 workshops.
Present resolution, quality, and cost metrics to the vice president of product monthly.

Theo Sandberg

(917) 555-0138 ◇ New York, NY

theo.sandberg@example.com ◇ linkedin.com/in/theo-sandberg ◇ theosandberg.com

Objective

AI prompt engineer with 3 years building and optimising prompts and agent workflows for enterprise applications, currently on the applied AI team of a legal technology company where language model features process 900,000 documents a month for 400 law firm customers. Designed the prompt and workflow for a contract review feature that reaches 94% agreement with attorney reviewers on a 2,000-clause benchmark, cut per-document model cost 52% through prompt restructuring and model routing, and built 30 evaluation sets used by 5 feature teams. Comes from a legal operations background and writes prompts the way a good associate writes a memo: precise, structured, and checked against the source.

Education

B.S. in Information Science, Hudson Crown University 2016 – 2020

Coursework in Natural Language Processing, Information Retrieval, Data Analysis, Human-Computer Interaction, and Programming in Python.

Certificate in Applied Generative AI and Legal Operations Certification, Empire Line University Extension and Corporate Legal Operations Consortium 2021 – 2023

Also completed a retrieval-augmented generation engineering course in 2024 and an AI evaluation methods course in 2025.

Skills

AI Prompt Engineering

Prompt design for extraction, classification, summarisation, and drafting tasks, agent and multi-step workflow prompting, tool calling and structured output design, retrieval-augmented generation prompts with citation, model routing and cost optimisation, prompt injection defence

Evaluation and Quality

Benchmark and evaluation set construction, expert labelling coordination, accuracy and agreement measurement, error taxonomy and failure analysis, regression testing on prompt changes, model comparison studies, cost and latency profiling

Domain and Collaboration

Legal document structure and contract review knowledge, working with attorneys as subject matter experts, product and engineering collaboration, documentation and prompt style guides, customer feedback analysis, training internal teams

Tools & Platforms

Python with pandas, Claude and other model APIs, LangChain and LlamaIndex, Braintrust, Weights and Biases, Jupyter, SQL, Git, Label Studio, Jira

Experience

AI Prompt Engineer 02/2023 – Present

Hudson Crown Legal Technology New York, NY

Design and optimise prompts and agent workflows for language model features processing 900,000 documents a month for 400 law firm customers, on an applied AI team of 8.
Designed the prompt and workflow for a contract review feature that reaches 94% agreement with attorney reviewers on a 2,000-clause benchmark, up from 71% at the first prototype, and flags risk clauses with a 96% recall.
Cut per-document model cost 52% through prompt restructuring, output schema tightening, and routing 60% of documents to a smaller model with 0 measured quality loss, and built 30 evaluation sets used by 5 feature teams.

Legal Operations Analyst and AI Specialist 08/2020 – 01/2023

Empire Line Corporate Legal Services New York, NY

Supported a legal operations team managing 6,000 contracts a year, then led the first language model pilots for contract summarisation and clause extraction over the final 14 months.
Built prompts and an evaluation process for a clause extraction pilot that reached 89% accuracy on 800 contracts and cut first-pass review time 60%.
Built a 40-page prompt and usage guide adopted by 30 legal staff with a 4.6 out of 5 usefulness rating.

Information Science Intern 06/2019 – 08/2019

Empire Line Corporate Legal Services New York, NY

Built a contract metadata taxonomy of 60 fields during a 12-week internship, applied to 2,000 contracts with a 97% tagging accuracy at review.
Wrote Python scripts that automated 4 weekly reports, saving the team about 6 hours a week.
Received a full-time offer at the end of the 12-week internship based on the taxonomy and automation work.

Projects

Contract Review Agent Workflow. Designed a 4-step agent workflow for contract review with a clause segmentation prompt, a risk classification prompt with a 30-category taxonomy developed with 6 attorneys, a redline suggestion prompt with citation to the customer playbook, and a confidence-gated escalation, validated on a 2,000-clause benchmark through 60 iterations, which reached 94% attorney agreement, 96% risk clause recall, and is used on 400,000 documents a month.

Model Cost Optimisation Programme. Ran a cost programme with prompt compression that cut input tokens 35%, tightened output schemas, a routing classifier sending 60% of documents to a smaller model, and caching of repeated context, validated against 30 evaluation sets, which cut per-document model cost 52%, saving about $1.4M a year, with 0 measured quality loss.

Evaluation Set Library. Built 30 evaluation sets across 5 feature areas with 6,000 expert-labelled examples, labelling guidelines reviewed by attorneys, and an inter-annotator agreement above 0.85 on each set, with a regression runner on every prompt change, which cut prompt regressions reaching customers to 0 in 18 months and is used by 5 feature teams.

Extra-Curricular Activities

Member of a legal technology community, presenting at 2 events on evaluation for legal AI.
Volunteer with a legal aid nonprofit, building a document intake assistant used by 20 staff in 2025.
Run about 20 miles a week and completed 2 marathons since 2023.

Leadership

Own the prompt review process and evaluation standard for an applied AI team of 8 and 5 feature teams.
Coordinate a panel of 6 attorney reviewers for labelling and benchmark validation.
Trained 25 engineers and product managers on prompt design and evaluation through 6 workshops.

Prompt Engineer Resume Example

AI Prompt Engineer Resume Example

More Resume Examples

Recommended Articles

Ready to Transform Your Job Search?

Products

Free Tools

Resources

For

Company

Developers