AI PhD Candidate at CiTIUS (USC). I build neuromorphic architectures — spiking (SNN/LIF) and continuous-time (CfC) networks — for clinical NLP and early-risk detection on social media. I pair that research with full-stack engineering, from distributed Rust runtimes to P2P apps and interactive web interfaces.
Current line: moving beyond hard top-K routing toward biologically-motivated gating (spiking & continuous-time) for interpretable, time-resolved mental-health modeling.
|
Neuromorphic symptom routing (LIF spikes / CfC) for early depression detection on social media. Replaces collapse-prone top-K routing with per-user, per-symptom gating anchored to BDI-II — interpretability and accuracy above the RNN ceiling (F1 .617 on eRisk 2022).
|
A neuromorphic topic model for time-resolved theme evolution and clinical grounding in early-risk text collections. Brings spiking dynamics to temporal topic modeling so themes emerge — and fade — over a user's timeline.
|
|
Pipeline-and-filter framework with hash-based caching for reproducible, modular ML experiments. Compute once, reuse everywhere — across runs and across teams. Published in SoftwareX (2026). |
|
|
An exploratory temporal vector database in Rust that treats time as a native geometric dimension of embedding space — storing vectors as trajectories, not snapshots (velocity, drift, causal retrieval).
|
From CHATTY (a tool to build & evaluate customizable LLM assistants) to Chatty the Lab: multi-model comparison and a visual Agent Studio with parallel graph execution, on a Rust + SvelteKit stack.
|
Published
- Exploiting topic analysis models to explore psychological dimensions in social media data — Scientific Reports (2026, Q1). Read →
- A study of word embedding models for measuring topic coherence — Knowledge and Information Systems (2026, Q2). Read →
- LabChain: Enabling reproducible and modular scientific experiments in Python — SoftwareX (2026). Read →
- Analyzing Gambling Addictions: A Spanish Corpus for Understanding Pathological Behavior — Findings of the ACL: EMNLP 2025. Read →
- Temporal Word Embeddings for Early Detection of Psychological Disorders on Social Media — Journal of Healthcare Informatics Research (2025, Q1). Read →
- Comparison of Clustering Algorithms for Knowledge Discovery in Social Media Publications: A Case Study of Mental Health Analysis — SEPLN / Procesamiento del Lenguaje Natural (2024). Read →
- Temporal Word Embeddings for Early Detection of Signs of Depression — CIRCLE 2022 (CEUR Vol-3178). Read →
Accepted — to appear
- Title TBC — Venue TBC (accepted, in press).
- Title TBC — Venue TBC (accepted, in press).
Full list & metrics on Google Scholar.