Experience
Research, industry, and internship roles — from doctoral research on data-center scheduling to six years in ad operations analytics.
Nov 2023 — Present
Ph.D. Candidate in Information & Communication Technology
Sapienza University of Rome, Italy — Advisor: Prof. Andrea Baiocchi (DIET).
- Performance analysis and scheduling in large-scale computing systems, from CPU data-center clusters to GPU-based LLM inference serving.
- Practical dispatching policies that work without knowing job sizes in advance, validated on Google Borg and Alibaba production traces with simulation and queueing models.
- Benchmarking how several large language models share one GPU server — offloading, preemption, and cache migration across multi-GPU platforms.
- Mentoring M.Sc. Data Science students (two graduated; one ongoing on energy efficiency in large-scale clusters).
Feb 2026 — Jul 2026
Visiting Researcher, IMDEA Networks Institute
Leganés, Spain — Supervisor: Prof. Marco Ajmone Marsan.
- How synchronization constraints at job start/end affect delay (response time, slowdown) and power consumption of data-center scheduling strategies on Google Borg workloads.
- Multiserver dispatching policies for heterogeneous clusters benchmarked on real production traces.
- Multi-resource-aware scheduling to reduce wasted resources and response times in GPU-accelerated multi-tenant environments.
Jan 2019 — Jan 2025 · San Francisco, CA
Operations Analyst → Lead Ad Operations Analyst, Compulse (formerly ZypTv)
Six years leading ad operations at a US ad-tech company.
- Built and maintained automated reporting pipelines and analytical infrastructure integrating data across multiple advertising platforms.
- Analyzed large-scale impression data to forecast trends, resolve anomalies, and optimize CTR, CPM, VCR, and conversion metrics.
- Led a team of operations analysts; owned delivery end-to-end and reported strategic progress to executive leadership.
Sep 2021 — Apr 2022 · Rome, Italy
Data Analyst Intern, UNICEF Italia
- ML models for donor behavior prediction on 21M+ euro donation data.
- PowerBI dashboards for real-time monitoring; end-to-end data lifecycle management.
Jun 2018 — Aug 2018 · Izmir, Turkey
Data Science Intern, Hugo Boss AG
- Forecasting industrial electricity consumption from high-frequency sensor data (ARIMA, Prophet, LSTM).
Education
Nov 2023 — Present · defense expected Jan 2027
Ph.D. in Information and Communication Technology
Sapienza University of Rome, Italy — Advisor: Prof. Andrea Baiocchi (DIET). Dissertation: Performance analysis and scheduling in large-scale computing systems, from CPU data-center clusters to GPU-based LLM inference serving.
Oct 2020 — Mar 2023 · 110/110
M.Sc. in Data Science
Sapienza University of Rome, Italy — Thesis: Job and Task Workload Generator Based on Google Data Center Measurements for Scheduling Optimization.
Sep 2015 — Jun 2019 · 3.76/4.0
B.Sc. in Industrial Engineering
Yasar University, Izmir, Turkey.