Sandra Acevedo
Independent Researcher | Cultural Analysis + Data Systems
I work at the intersection of Latin American critical theory, data annotation, and natural language processing,
with a focus on Spanish-language representation, culturally situated evaluation, and the design of clearer,
more accountable knowledge systems.
Current work: NLPgap research on culturally situated NLP data curation and evaluation. In parallel, I build
small, inspectable artifacts (schema + annotated entries + QA logic) that make the method visible.
Samples
·
Repo
¿Quieres leerme en español?
Fine Arts Theory + Cultural Management
Universidad de Chile | Conceptual grounding in institutions, cultural mediation, and the politics of public access.
- Focused on how cultural institutions mediate public access to artistic heritage and reproduce hegemonies through institutional practice.
- Built the conceptual backbone that later became legible again inside technical systems and search evaluation environments.
What this proves: I bring a coherent intellectual backbone to applied work. Theory is not decoration here, but a diagnostic instrument for reading systems of authority, representation, and exclusion.
Transantiago
2008–2011 | Field-facing work at the interface of users, operators, and a large public transit system.
- On-the-ground interaction with users and operators working with operational technology.
- Observed recurring failure modes where system design meets real-world constraints: access, comprehension, friction, and error recovery.
- Built early “quality” instincts through lived service breakdowns and repair, not abstract theory.
What this proves: I learned evaluation inside a high-stakes public system: real users, real constraints, real consequences.
Web Search Evaluation
2016–2021 | Quality evaluation for a major search engine, applying editorial criteria and relevance judgments at scale.
- Evaluated nearly one million search queries while maintaining top performance metrics over a five-year period.
- Identified recurring patterns where supposedly neutral technical criteria encoded editorial bias and reproduced cultural hierarchies.
- Developed durable judgment practices around quality, authority, contextual relevance, and guideline-based consistency.
What this proves: I have hands-on experience inside a large-scale ranking and evaluation ecosystem, with a trained eye for how “quality” is constructed, enforced, and mis-specified.
Data + SQL Literacy
2024 | Formal technical grounding to connect lived experience in data work with analytical tooling and cross-functional collaboration.
- Completed Google Data Analytics and SQL-focused training to strengthen fluency in datasets, structure, and analytical workflows.
- Expanded the ability to reason about data pipelines and constraints in ways that support work with technical teams.
- Used technical upskilling to sharpen, not replace, critical interpretation and methodological clarity.
What this proves: I am not only commenting on systems from the outside. I can read their structure, understand their constraints, and work across the boundary between interpretation and implementation.
Independent Research
May 2025–present | Culturally situated NLP, Spanish-language representation, and data curation frameworks informed by Latin American critical cultural theory.
- Developing a research line focused on how cultural power structures appear in technical pipelines, evaluation criteria, and editorial defaults.
- Synthesizing Latin American cultural studies and contemporary NLP research into practical, inspectable approaches for curation and assessment.
- Building public-facing writing and research artifacts designed to be legible across humanities, education, and technical audiences.
What this proves: I can translate theory into operational framing for data work, and translate technical constraints back into cultural analysis without turning either into empty rhetoric.
Backbone Question
How do we build culturally situated AI systems when the pipeline itself is designed from dominant cultural frameworks?
Approach
My work is interdisciplinary by necessity. I combine Latin American critical traditions, cultural analysis,
editorial judgment, and NLP research to produce frameworks that can be implemented, evaluated, and argued for with rigor.
I work in Spanish and English, and I treat bilingualism as an epistemic advantage when auditing Anglo-centric defaults
and building more context-aware forms of evaluation.
What I Offer
- Research & Analysis: cultural analysis, AI evaluation framing, institutional critique, and applied inquiry across humanities and technical domains
- Quality & Curation: data annotation QA, guideline interpretation, editorial judgment at scale, and bias-sensitive evaluation
- Information Work: metadata thinking, taxonomy and knowledge-organization logic, and bridge work between humanities, education, and data systems
- Writing & Communication: academic and technical writing, public scholarship, bilingual content, and clear adaptations of complex ideas across audiences
Location: United States
Languages: Spanish (native), English (fluent)
Email
·
LinkedIn