PhD student in the Graduate Program in Electronic Engineering and Computing (Telecommunications area) at ITA — Instituto Tecnológico de Aeronáutica, São José dos Campos, Brazil. Advised by Prof. Sarah Negreiros de Carvalho Leite.
My research is in NILM (Non-Intrusive Load Monitoring) — disaggregating a household's total electricity signal into individual appliance loads using machine learning — with a long-term goal of embedded deployment in smart meters.
- 🎓 PhD started February 2026 · expected completion ~Dec 2029 / Feb 2030
- 🔬 Background in stochastic processes and statistical signal processing (WSS processes, PSD, blind source separation)
- 🧠 Applies deep learning architectures to NILM disaggregation
- 📍 São José dos Campos, Brazil
| Stage | Approach | Status |
|---|---|---|
| 1 | Seq2Seq RNNs (LSTM, GRU, BiLSTM, BiGRU, SimpleRNN) with sparsity-weighted adaptive loss and a gated decoder | Master's thesis, REDD dataset — manuscript under review, IEEE Transactions journal |
| 2 | Transformer-based disaggregation — shared encoder with appliance-specific cross-attention decoder heads | Current PhD work, REDD dataset — IEEE-format paper drafted |
| 3 | Probabilistic unsupervised adaptation (F-HDPHSMM) | In progress |
Evaluated with F1-score, MAE, and Signal Aggregate Error (SAE). Also exploring zero-/few-shot LLM-based disaggregation as a side track.
- Modelo de Otimização Aplicada ao Despacho de Veículos de Emergência para Atendimento aos Clientes com Interrupção de Energia Elétrica — undergraduate IC paper, advised by Prof. Mário Mestria. Link
- Otimização de Trajetórias através de Caminhos Mínimos para a Locomoção de Cadeira de Rodas Robótica — XI Simpósio Brasileiro de Automação Inteligente (SBAI 2013), Fortaleza-CE, advised by Prof. Mário Mestria. Link
- Seq2Seq RNN NILM manuscript (with Marcelo S. de Castro and Sarah N. de Carvalho Leite) — under review, IEEE Transactions journal
- Transformer-based NILM disaggregation paper — IEEE format, in preparation
Full record on Google Scholar