Helsinki, Uusimaa, Finland
Data-driven AI/ML researcher with 6+ years of experience in resource-optimization and intelligent automation for complex systems. Currently working on network observability using agentic AI frameworks, designing AI agents and RAG pipelines for automated root-cause analysis, and chaos-engineering-driven resilience testing. Specialized in statistical modeling, time-series forecasting, AI/ML, and linear/integer optimization for resource allocation, scheduling, and routing in wireless networks. Google Scholar Profile: https://scholar.google.com/citations?user=bZC5OXQAAAAJ&hl=fi
1. Defined problem statements, designed hypotheses, and built end-to-end ML and optimization pipelines for forecasting, scheduling, and routing. Produced reproducible code and experiments, published in peer-reviewed journals, and collaborated with international teams, including research visits to the University of Málaga (ERTIS Group), Spain, and TUAT Umebayashi Lab in Tokyo, Japan. 2. Designed and implemented EV-Charging Optimization System — real + GAN-synthetic data and Bayesian NN arrival prediction + MILP/heuristic scheduling algorithms to dynamically assign charging slots for minimal energy load and wait times. 3. Designed and implemented Cost-Aware EV Mega-Station Control System — AE-LSTM (BNN)‑driven arrival forecasting + M(t)/M/s(t) queuing + non‑linear cost optimization to reduce waiting time by ≈30% and lower total cost by ≈12–25% through tolerable waiting-time thresholds, while minimizing charging sessions and grid reservation cost with maintained QoS. 4. Designed and implemented Uncertainty-aware EV Delivery-Routing for Battery-as-a-Service (BaaS) Systems — BNN-based road-traffic speed forecasting + MILP/heuristic routing algorithms to ensure reliable battery delivery and minimize power-outage risk in cellular communication networks.
1. Defined problem statements, designed hypotheses, and built end-to-end ML and optimization pipelines for forecasting and allocation. Delivered reproducible code and experiments, published in peer-reviewed journals, and supported academic activities as a teaching assistant and technical supervisor for graduate research. 2. Developed WLAN Traffic Forecasting Pipeline — using real data from 470 Aps, data classification, filtering, aggregating, correlation and stationary analysis, temporal and spatial data modeling, and CNN-LSTM development for proactive resource allocation in an enterprise network. 3. Enhanced Wireless Network Data Forecasting Engine — combining physical‑layer and network‑layer data to enhance channel‑demand prediction in enterprise WLANs. 4. Designed and implemented Uncertainty-aware Interference Forecasting & Secondary Users Allocation — multinomial data modeling and LSTM/Transformer + Monte‑Carlo uncertainty estimation to boost users’ throughput by ≥52% while proactively constraining transmissions to protect radar operations in a radar-spectrum sharing system. 5. Enhanced Radar-Spectrum Sharing Engine — GRU uncertainty-aware model to achieve ≥13% faster prediction and strengthen radar protection under sensitive operating thresholds.
As a Visiting Researcher on the EVOLVE project - conducted research on optimizing ML-based routing algorithms for EVs in collaboration with Ertis Research Group.
As a Visiting Researcher on the EVOLVE project - conducted research on optimizing ML-based scheduling algorithms for EV charging stations in collaboration with TUAT's Umebayashi Lab.
Integrated MATLAB-based DSP modules into a software-defined communication system, enabling end-to-end text-message encoding and decoding between access points.