ScalaF: Functional Refactoring for Scala
Built a static analysis engine using Scalameta to detect imperative anti-patterns and safely refactor them into functional and parallel Scala code. Presented at Scala Workshop 2025 (SPLASH/ICFP).
Applied research for scalable and reliable AI systems.
I focus on practical research where algorithmic ideas translate into usable systems, with strong attention to performance, safety, and maintainability.
Ransomware Early Detection with Reinforcement Learning
Framed early malware detection as an optimal stopping problem and used PPO with a BiLSTM policy on API call sequences, improving early-stage detection while reducing the number of observed calls.
Speech-Driven Customer Service Assistant
Developed an end-to-end STT -> RAG -> TTS architecture for voice-first support experiences, including constrained decoding strategies to improve response quality and speech naturalness.
Localization and Deterministic Testing Systems
Engineered software modules for high-precision localization and deterministic software-in-the-loop validation, reinforcing reliability expectations for safety-critical environments.