Sofia Hirschmann
Education
Ph.D. in Computer Science, Northeastern University
Sept. 2023 - May 2028Awards & Honors: Khoury Distinguished Fellow
Research Focus: Human-AI collaboration and cognitive modeling
B.S. in Data Science and Cognitive Psychology, Simmons University
Sept. 2019 - May 2023Awards & Honors: Summa Cum Laude, Honors Program
Minors: Computer Science and Spanish Literature
Professional Experience
Doctoral Researcher, Northeastern University
Sept. 2023 - Present- Modeled human cognition in cooperative and adversarial AI settings.
- Built PsychSim-based POMDP and Theory of Mind simulations in Python encoding bounded rationality and cognitive biases (loss aversion, sunk cost, base-rate neglect) to predict network attacker action sequences, validated on red-team traces and synthetic scenarios.
- Designed large-scale web experiments (1,000+ participants) to study adaptation in human-AI collaboration, using Bayesian inference over human capability and intent to adapt agent behavior.
- Analyzed behavioral and survey data to derive design guidelines for transparency, coordination, and role signaling in human-robot teams.
Applied AI/ML Fellow, Advisor360
Sept. 2025 - Present- Built an LLM + RAG Slack triage system to reduce ticket time-to-resolution and overall ticket volume across production and non-production workflows.
- Translated developer interviews into a routing policy and built an internal service-to-team mapping to power an auto-assignment agent achieving ~80% routing accuracy.
- Developed an interactive analytics and retrieval agent that summarizes recurring issues and bottlenecks and recommends resolutions by embedding-based matching to historically resolved tickets.
Data Science Researcher, MITRE
May 2022 - June 2023- Automated the National Imagery Interpretability Rating Scale (NIIRS) evaluation workflow in Python, replacing manual scoring with a reproducible ML pipeline.
- Built a predictive model of tracker performance using a feature set grounded in motion-imagery quality factors and validated on held-out sequences; peer-reviewed results published in IEEE AIPR.
Research Assistant, Simmons University
July 2022 - June 2023- Annotated 500+ utterances with PoLaR and supported a prosody-to-intent pipeline in R by extracting prosodic features (pitch, intensity, duration) and training supervised models.
- Ran feature ablations to quantify contributions of rhythm, stress, and intonation to speaker intent understanding.
Selected Papers & Presentations
- Guarding Against Malicious Biased Threats (GAMBIT): Co-authored a cognitive-informed cyber defense framework that models attacker decision-making biases to improve attack prediction.
- Theory of Mind in Human-Robot Task Collaboration: Presented at CHI ToM Workshop (May 2024); investigated how predictive reasoning and action clarity influence coordination in turn-taking games.
- Relationship Between Tracking Performance and Video Quality: First-authored and published at IEEE AIPR (Oct 2022); developed feature-based models linking video quality to automated object tracking.
- CCSCNE Conference (April 2023): Presented poster on machine learning methods to extract meaning from prosody in language.
In the Works
- Preparing paper/journal regarding adversarial and collaborative Theory of Mind.
- Preparing paper on cognitive biases decision making mechanics in virtual escape room.