Looks like you have JavaScript disabled. For the full Twine experience, you will need to re-enable it.

I am a passionate data scientist and AI engineer with a keen interest in leveraging artificial intelligence to create impactful solutions. My journey has been defined by my commitment to continuous learning and exploration in the fields of data science and AI, and I enjoy collaborating with like-minded individuals to push the boundaries of technology.…

Sarah Rosaria Dias Barreto

Data Scientist, Developer, AI Engineer, +2





I am a passionate data scientist and AI engineer with a keen interest in leveraging artificial intelligence to create impactful solutions. My journey has been defined by my commitment to continuous learning and exploration in the fields of data science and AI, and I enjoy collaborating with like-minded individuals to push the boundaries of technology.…

Available to hire

I am a passionate data scientist and AI engineer with a keen interest in leveraging artificial intelligence to create impactful solutions. My journey has been defined by my commitment to continuous learning and exploration in the fields of data science and AI, and I enjoy collaborating with like-minded individuals to push the boundaries of technology.

Skills

Experience Level

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Intermediate

Intermediate

Intermediate

Language

English

Fluent

Spanish; Castilian

Intermediate

Work Experience

Research Scientist (Advanced Analytics) at Kelly School of Business

March 1, 2025 - June 7, 2024

Directed longitudinal survival analysis on 1,200+ entrepreneurs using multiverse modeling (R), uncovering key health and business risk factors. Analyzed 30+ high-dimensional behavioral and health attributes to identify drivers of entrepreneurial mortality and resilience.

Data Scientist at Project 990

March 1, 2024 - June 7, 2024

Automated Alteryx workflows to filter and process 500K+ congregation records, reducing data preparation time by 38% and boosting data quality scores by 19%. Conducted causal discovery with placebo testing, filtering 50% of models to improve causal effect estimation on education outcomes. Deployed zero-shot RoBERTa classifiers (NLP) for 500,000+ mission statements; accelerated inference time by 40% via DeepSpeed optimization. Engineered five scalable ETL pipelines, increasing data throughput by 35% and reducing integration errors by 23%. Developed 10+ Tableau dashboards to distill complex data into strategic insights, accelerating decision cycles by 36%.

Data Science Intern at Public Budgeting and Finance

May 1, 2024 - March 1, 2025

Created an automated email service and a website to advertise the journal, leading to a 11% increase in research engagement. Analyzed performance metrics post-deployment of both frontend and backend on a single Heroku dyno, which reduced hosting costs by 30% and improved API call response times by 22%, enhancing user experience. Curated a repository of over 600 researchers, streamlining collaboration and boosting project matching speed by 25%. Leveraged pretrained BERT models and cosine similarity to identify matching articles, improving matching accuracy by 85% and reducing researchers’ article search time by 42%.

Data Science Research Assistant at O’Neil School of Public Affairs

September 1, 2024 - January 31, 2025

Implemented rigorous A/B testing protocols to assess and compare regex matching with fuzzy logic for efficient scraping, leading to improved extraction rates of financial data from unstructured reports by 30%. Automated document parsing pipelines in Python, achieving 50% accuracy across highly variable financial documents.

Data Analyst Intern at Indiana University

March 1, 2024 - May 31, 2024

Engineered scalable data pipelines (ETL) for 5,000+ record datasets, reducing missingness by 55% through advanced imputation methods. Conducted data validation and quality checks, achieving 99% verified accuracy for research-grade datasets.

AI Engineer Intern at Aynak

August 1, 2025 - Present

Finetuned and deployed DeepFilterNet for speech enhancement, achieving +0.3 PESQ and +10% STOI compared to SEGAN. Reduced computational cost by 70% and maintained sub-20 ms latency, enabling deployment on smart glasses.

Data Scientist at Project 990 USA

March 1, 2024 - August 1, 2025

Leveraged RoBERTa for zero-shot text classification on 10,000+ mission statements per batch, extracting features from 150+ faith- and gender-based keywords. Optimized inference with DeepSpeed, reducing classification time by 40%.

Founding AI Engineer at Aynak

August 1, 2025 - Present

Led core AI development, engineering a low-latency speech-enhancement pipeline that cut computational load by 66% while achieving stable 10ms real-time performance. Refactored and stabilized 3,000+ lines of DSP/C++ inference code, improving throughput and eliminating failures.