GenAI Developer @ Ericsson · Computer Science @ Concordia University
I work on applied machine learning and on LLM systems: PyTorch models, evaluation, APIs, and data pipelines.
Python · Java · PyTorch · TensorFlow · scikit-learn · Computer Vision · NumPy · pandas · FastAPI · Spring Boot · Docker · PostgreSQL
At Ericsson I work on generative AI engineering and AI-enabled software workflows.
| Project | Problem | What I built |
|---|---|---|
| HalluciScope | RAG answers that look right but are not grounded in the source | Claim-level ensemble: NLI, self-consistency, semantic similarity. FastAPI + React, FAISS, Ollama. Per-claim hallucination score + HaluEval ablation. |
| Surgical workflow | Joint phase recognition and tool tracking in laparoscopic video | Multi-task MS-TCN on frozen ResNet-50 features, plus a correlation loss for impossible tool-phase pairs. Four-model comparison, 64 tests, Cholec80 subset. |
| NOVA | Predict node outages from telemetry and syslog | Ingest → windowed features → supervised MLP + autoencoder. Public-data reimplementation of internship work. |
| Distributed Task API | Task API that survives duplicate writes and concurrent updates | Java 21 / Spring Boot: idempotency keys, optimistic locking, Flyway, Testcontainers, Docker Compose. |
Computer vision and time-series ML, plus GenAI systems at Ericsson.



