Louis Raison
Cambridge, Massachusetts, United States
717 followers
500+ connections
About
I'm a Boston based data scientist and engineer, leveraging cutting-edge optimization…
Activity
717 followers
-
Louis Raison shared thisZipline is celebrating one million deliveries today! That’s one million packages of lifesaving blood, critical vaccines, last-minute groceries, and more. I feel proud to be contributing to this project that’s helping communities all over the world access the things they need quickly, safely, and sustainably.Louis Raison shared thisWe’re thrilled to announce our 1 millionth delivery today! This major milestone wouldn’t be possible without the tireless work of the entire Zipline team and our incredible partners all over the world. One million deliveries completed. Millions more ahead. #Zipline #DroneDelivery #MakingAnImpact https://lnkd.in/gePn8Xdd
-
Louis Raison shared thisI'm happy to announce that I have joined the Data Team at Zipline as a Data Scientist. Zipline is building a faster and more equitable delivery system using autonomous drones and I’m so excited to be a part of it. Those last three years at AlphaRoute have been amazing, full of impact and great learning experiences with my team. Special thanks to Mohammad Ghane-Ezabadi and John Hanlon for being insightful mentors, and Chelsea Alford, Kailyn Bryk, Saksham Soni, and Raphaël Maillard for being very smart, caring, and hardworking coworkers that I have had a pleasure learning and building with! I'm really looking forward to those next years that will for sure offer their share of challenges and achievements. #Zipline
-
Louis Raison shared thisOur team at Dynamic Ideas Routing is expanding, building better transportation networks for schools and individuals! Join us!Louis Raison shared thisCome works with us! Our team is looking for a Data Analyst and a Data Scientist to help us transform the world of mobility. #datascience #machinelearning #datascientist #analytics https://lnkd.in/dpZRdSE
-
Louis Raison shared thisI am proud and delighted of celebrating my graduation today from the MBAn program at MIT Sloan and the nomination of the capstone project Yanchunni (Stella) Guo and I conducted with BMW Group as one of the top two projects for the MBAn this year! Our work on vehicle noise classification was one of the most rewarding times in my life, thanks to the great support of Maka and Matteo from the BMW team. Working remotely with one person in Singapore, BMW in Germany and me in the US was certainly hard but still yielded exceptional results. Starting now, I will be working on school buses network optimization in the US with Dynamic Ideas Routing, a project that I started contributing to as a research assistant this year and that means a lot to me. Thanks to everyone that helped us during this project and congratulations to all of my fellow MBAn graduates! I look forward to seeing you again in the future.
-
Louis Raison shared thisIpyaggrid : Harnessing the power of ag-Grid in Jupyter. Beautiful and efficient data display in notebooks with zero JS coding. https://lnkd.in/d7Bdraj
-
Louis Raison reacted on thisLouis Raison reacted on thisI am excited and honored to announce that I am stepping into an official leadership role on my team! I now manage the P2 Analytics Engineering team, a small but mighty crew building the foundation of P2 business intelligence and analytics at Zipline. Like many of us, my job requirements have evolved significantly over the past year. I transitioned from primarily "writing SQL to work with data," to "using AI to work with data," and now to "planning and implementing our platform strategy to empower previously non-technical stakeholders to use AI to work with data." Phew! Additionally, I will be hiring in the coming months! The job description is somewhat flexible, as we are continuously learning new skills to better serve the business. If you have a solid foundation in software, data, or analytics engineering, and are interested in solving new and exciting problems each day, please reach out!
-
Louis Raison reacted on thisLouis Raison reacted on thisReflecting on my time at Zipline, it’s incredible how much I’ve learned in just a few short months. Spring in the Bay Area has been unreal, and I’m not quite ready for it to end. That’s why I’m proud to say 𝐈’𝐦 𝐧𝐨𝐭 𝐝𝐨𝐧𝐞 𝐲𝐞𝐭, as I’ll be returning to Zipline in South City this summer to continue growing in data science. Although I wasn’t a new recruit, I’m grateful to Anders Nottrott, Winter Pickett, and Michelle Park for a seamless extension process and the opportunity to stay. Excited to keep building for Zipline’s autonomous logistics system as it continues to scale.
Experience
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
Education
-
MIT Sloan School of Management
3.9/4
-
- Ranked the best capstone project of the year
- Relevant Coursework: Machine Learning, Optimization Methods, Supply Chain, Operations Management
- Projects:
1. Predicted the demand and optimizing the allocation of taxis in different districts of NYC to reduce the number of taxis needed
2. Used Robust Optimization methods to optimize deployment of Police patrols in Boston
3. Predicted customer orders on a location-to-warehouse lane to consolidate them for Lineage…- Ranked the best capstone project of the year
- Relevant Coursework: Machine Learning, Optimization Methods, Supply Chain, Operations Management
- Projects:
1. Predicted the demand and optimizing the allocation of taxis in different districts of NYC to reduce the number of taxis needed
2. Used Robust Optimization methods to optimize deployment of Police patrols in Boston
3. Predicted customer orders on a location-to-warehouse lane to consolidate them for Lineage Logistics -
-
-
-
-
Licenses & Certifications
-
-
SJB1-2.101x: Steps in Japanese for Beginners1 Part2
edX
Issued ExpiresCredential ID 970866a792fe4fabadf163478041674b
Projects
-
IpyAgGrid
See projectIpyAgGrid is a jupyter tool designed to enable easy access to basic ag-Grid features to non-JavaScript knowledgeable users, but also allows unlimited customization through a full access to the ag-Grid API.
-
IPyHC
-
- Data-visualization package development for Jupyter notebook, based on the HighCharts graph library.
- Open-source developmentOther creators -
Languages
-
English
Native or bilingual proficiency
-
Spanish
Limited working proficiency
-
French
Native or bilingual proficiency
Other similar profiles
Explore more posts
-
Youssef Hosni
A lightweight VLM outperforms much larger models in OCR. With just 1.7B parameters, dots.ocr sets a new state of the art on OmniDocBench, showing that strong document understanding doesn’t require massive models or costly APIs. Most current OCR leaders rely on very large models (72B+) or paid services like GPT-4o. dots.ocr takes a different approach. Built on a compact 1.7B LLM foundation, it outperforms heavyweight models such as Qwen2-VL-72B and GPT-4o across key benchmarks. Why it stands out: ➡️ Unified architecture: Processes text, tables, formulas, and layout in a single pass ➡️ High precision: Achieves an overall edit distance of 0.125 on ➡️ OmniDocBench (lower is better), ahead of GPT-4o at 0.233 ➡️ Production-ready: Integrated with vLLM for high-throughput deployment Performance highlights: ➡️ Table recognition: 88.6 TEDS score (vs. 72.0 for GPT-4o) ➡️ Layout detection: 0.930 F1 score, outperforming DocLayout-YOLO ➡️ Inference speed: Faster than larger foundation models thanks to its compact size You can deploy dots.ocr locally using vLLM or Hugging Face. The GitHub repo and huggingface page are in the comments.
3 Comments -
Latent Scholar LLC
Gaussian processes are theoretically compelling but computationally brutal at scale. Can sparse inducing points and stochastic variational inference actually make non-conjugate GP models practical for large datasets? We asked Google Gemini 3.1 Pro to work through the full methodology — from variational lower bounds to Gauss-Hermite quadrature. If you work in Bayesian ML, this one needs your critical eye. #MachineLearning #BayesianInference #GaussianProcesses #VariationalInference #AI
1 Comment -
Towards Data Science
Struggling with TorchScript's limitations when deploying PyTorch models? Learn how torch.export, a new API highlighted in Chaim Rand's article, overcomes these hurdles by successfully tracing models with dynamic shapes and conditional logic, offering a more robust deployment solution.
Explore top content on LinkedIn
Find curated posts and insights for relevant topics all in one place.
View top content