Adrian C. Prelipcean

As a founder: - took the company from an "Uber for deliveries" idea lacking products and market fit to one of the leading logistics companies in Sweden - scaled up Airmee from 2 people to roughly 100 people - led everything tech and research related - board member (until 2020) - led the technical due diligence process of all investment rounds As an engineer: - designed the system architecture and built the prototypes for all core products and services (roughly 15 different products), a system that scaled for millions of deliveries in 3 different countries with negligible hosting cost - responsible for team structure and management, one on ones, hiring processes, onboarding new engineers - designed, built, and led the team for integration solutions with external partners that range from stores operating at a local scale to international enterprises (e.g., Amazon, H&M, Shein, etc.) - grew the Airmee tech team from me being the sole developer to 20 engineers - responsible for AWS cloud administration (until 2023), PostgreSQL database administration (until 2022) and establishing the engineering best practices for frequent and stable releases (CI/CD) As a data scientists: - designed, built, and led the team for real time and close to real time optimization services - built data ingestion pipelines that automatically generate the data and infrastructure necessary for high quality optimization in new regions - led the development of analytics processes for analyzing performance of models and optimization output by comparing model output to real world performance - built models for predicting the amount of time it would take a courier to hand over a delivery - designed processes that combine academic innovation with engineering reliability

• Proposed a new method for assessing the accuracy of classification done on continuous 2D phenomena. The method utilizes longest common sequences on top of an alphabet of activities to identify the magnitude of error across either dimension. • Proposed and implemented MEILI, a system that is used for collecting GPS trajectories from population subsets and automatically annotate the trajectories into travel diaries. For the annotation methodology, multiple machine learning and statistical methods were utilized such as: rule based systems, decision trees, random forests, neural networks, etc. • Implemented an algorithm that adjusts a smartphone’s location sampling frequency to optimize for equal time or equal distance units between consecutive GPS points. This decreased smartphone battery usage by up to 50% while maintaining high sampling quality. • Proposed various methods on comparing automatically generated travel diaries with manually declared travel diaries. This is a difficult problem because objective ground truth is not achievable, since either method has its shortcomings. • Used Weka to build a recommender system that allowed users to upload a set of papers or articles and get suggestions on relevant courses and teaching staff. The paper content was tokenized and the result was compared using different NLP methods with tokenized course and instructor information