Laurens Michielsen

Incoming Software Engineer Intern @ Palantir Technologies

About

I am a Machine Learning/AI engineer with a strong background in probabilistic modeling, computer vision, and large-scale data analysis. I bring hands-on experience in deploying ML systems in safety-critical and production environments. Currently, I am pursuing an MS in Computer Science (Machine Learning track) at Columbia University, after completing my BSc cum laude at TU Delft. My work sits at the intersection of machine learning, systems, and decision-making under uncertainty. As Chief Embedded & Autonomous Software at Formula Student Team Delft, I led the redesign and deployment of a fully autonomous (driverless) electric race car. I owned the technical decisions and system architecture across perception, SLAM, path planning, control, embedded software, telemetry, and sensor selection. I applied probabilistic state estimation, statistical analysis, and large-scale backtesting on noisy time-series data to improve localization accuracy and downstream autonomous stability. I also designed and tested LiDAR-camera fusion pipelines, led the integration of all the hardware, and led the team through simulation, track testing, and competition deployment. Alongside autonomous driving (AV), I have experience building scalable data infrastructure for computer-vision workflows. I also conducted ML robustness research, including large-scale experiments on object detectors under different label noise distributions and fine-tuning deep learning models on HPC systems. More recently, I have worked extensively with foundation models, building RAG pipelines, multi-agent LLM systems, and fine-tuned transformer models, with a focus on reliability, grounding, and evaluation. I am looking for AI/ML Engineer roles in autonomous vehicles, robotics or finance, where I can own complex ML systems end to end—from modeling and experimentation to deployment and monitoring. I am especially drawn to teams that value robustness, interpretability, and sound statistical reasoning, and that operate at the intersection of ML, systems, and real-world impact.

Country

United States

City

New York

Industry

Computer Software

Skill

Object Detection, Artificial Neural Networks, Convolutional Neural Networks (CNN), Statistics, Statistical Reporting, Statistical Analysis, High Performance Computing (HPC), Slurm Workload Manager, Research Skills, Design of Experiments (DOE), Experimental Design, Probability Theory, Probabilistic Models, Quantitative Analytics, InfluxDB, Marple, Raspberry Pi, Retrieval-Augmented Generation (RAG), RLHF, Autonomous Vehicles

Experience

Incoming Software Engineer Intern

Palantir Technologies

2026-3 - Present · 7 mos

New York City Metropolitan Area

Co-Founder

Columbia Autonomous Racing

2025-11 - Present · 11 mos

New York, New York, United States

Chief Embedded & Autonomous Software

Formula Student Team Delft

2024-8 - 2025-8 · 1 yr 1 mo

Delft, South Holland, Netherlands

I led the redesign, development and testing of the autonomous/driverless and embedded software stack for the electric race car, enabling the vehicle to race fully autonomously (AV). I led two teams, seven autonomous engineers and 1 embedded engineer, and was responsible for design decision, system architecture and validation. I also owned sensor selection and hardware integration for the driverless stack, including a LiDAR, camera, IMU, OGSS, and vehicle interfaces. I also designed and evaluated perception and SLAM pipelines within a ROS-based, real-time system focusing on robustness under noise, latency, and limited compute. I also developed embedded software for the 7 custom PCBs for the vehicle interfaces and to ensure rules compliance. I built a telemetry and logging system to support real-time monitoring and post-run analysis. This work resulted in 3rd place overall at Formula Student Germany, 2nd place Driverless Engineering Design at FS Czech (first attempt in 6 years), and the team's first-ever completed skidpad in its six-year driverless history with a time only 1.5s off the world record.

Researcher

Delft University of Technology

2024-4 - 2024-8 · 5 mos

Delft

Conducted large-scale research on object detection model robustness under synthetic label noise across datasets including PASCAL VOC, VisDrone, and medical imaging. Analyzed generalization degradation in fine-tuned detectors, showing YOLOv8’s superior stability over Faster R-CNN.

Software Engineer

Robovision

2023-8 - 2023-9 · 2 mos

Gent, Vlaanderen, België

Designed and deployed a scalable, event-driven data management system supporting industrial computer vision workflows. Scaled system capacity from 8000 files to production scale volumes by implementing automated change detection and API-based notifications. Improved efficiency by 15%. Recognized by supervisor for clear communication, initiative, and implementation quality.

Software Engineer

Vera Connect

2023-4 - 2023-8 · 5 mos

Rotterdam, Zuid-Holland, Nederland

Led a team of 5 peers to design and launch a B2C extension for an existing privacy-by-design contact sharing platform. Owned backend development using TypeScript, PostgreSQL, and Prisma, delivering APIs with 90%+ automated test coverage. Designed and implemented a responsive React + TypeScript UI, focusing on usability and clean user flows. Collaborated cross-functionally on product requirements, architecture decisions, and deployment.

Software Engineer

EXINmc - Prestatieverbetering voor Zorg en Bedrijfsleven

2022-7 - 2022-8 · 2 mos

Antwerp area

Education

Columbia University

Computer Science (Machine Learning)

2025-9 - 2026-12 · 1 yr 4 mos

Courses: LLM-based Generative AI, Deep Learning for Computer Vision, Analysis of Algorithms.

EPFL

Data Science and Intelligent Systems

2023-9 - 2024-2 · 6 mos

Courses: Markov Chains and Algorithmic Applications, Mathematics of Data: From Theory to Computation, Intelligent Agents, Intro to NLP, Intro to IT consulting.