uctp-metaheuristic-framework
Hybrid metaheuristic-RL framework for multi-objective university course timetabling. NSGA-II, PPO/DQN, multi-armed bandit operator selection, and CP-SAT decomposition repair.
Hi, I'm Krishna Acharya 👋
Optimization · Reinforcement Learning · Communication Systems · Embedded
github.com/krishna-jiBE in Electronics, Communication & Information Engineering — Thapathali Campus, IOE TU (2022–2026)
Based in Kathmandu, Nepal
About Me
- Published researcher with 3 papers spanning CNN-LSTM models, 6G cognitive radio, and assistive technology.
- Building multi-objective optimization engines using NSGA-II, MOEA/D, and reinforcement learning hybrids.
- Focused on 6G cognitive radio, dynamic spectrum access, and SDR signal processing.
- Working with reinforcement learning methods including DQN and PPO for scheduling, trading, and spectrum allocation.
- Exploring quantitative finance through RL trading agents, time series analysis, and 15+ NEPSE repositories.
- Building embedded systems with STM32, ESP32, and custom PCB design.
Featured Projects
nepse-rl-agent-rbs-experiment
Quantitative trading research platform for NEPSE with 7 rule-based strategies, LightGBM signal filtering, PPO RL agents, 15k+ LOC, and 99 tests.
vrp-solver
Multi-objective VRP solver using NSGA-II for distance, lateness, idle time, and fairness, with Kathmandu traffic modeling and demand forecasting.
automatic-rf-identification-cnn-lstm
CNN-LSTM hybrid for over-the-air automatic modulation classification using SDR. Published in JIEE 2025 with 93.48% accuracy across 10 modulation schemes.
Publications
Contact
For research, collaboration, or project discussions, reach me here.