From Cameroon to MIT: Geometry, Grace, and Growth

Few of the last five summers barely live in my memory, but summer 2024 stands apart. It was one of the most epic, transformative, and decisive periods of my life to date, reshaping me as an individual, a young academician, and practical Christian.

After months of relentlessly applying to over 150 internships — from Google to the T. Foundation, I'd grown used to the sting of rejection. Every day brought another polite "thanks for your time…" email. Then came the message from Professor Justin Solomon, director of the Geometric Data Processing Group at MIT, inviting me to join His Summer Geometry Initiative (SGI) research program. I was the only Cameroonian selected, and one of just two from African soil. That was the "Aha" moment! How do I explain this? — my feverish chills vanished instantly. I jumped from my table, shut the door, and lay down in tears and sweat for the next few minutes, thanking God for His Miracle! Then I ran to my sister next door, messaged my parents, wrote to my very good friend and mentor, Cleopatra E., who had been a previous fellow, then spent the next hour jumping around my room in disbelief and joy. Picture it!

What the SGI really is VS what it meant to me

Permit me to re-define what the SGI is in two ways. First from the perspective of a prospective applicant, and second as a fellow (myself, in this case).

As a prospective applicant, the SGI is a six-week paid summer research program introducing undergraduate and graduate students to the field of geometry processing. Geometry processing has a long history of breakthrough developments that have guided design of 3D tools for computer vision, additive manufacturing, scientific computing, and other disciplines. Algorithms for geometry processing combine ideas from disciplines including differential geometry, topology, physical simulation, statistics, and optimization.

Now, to me, the SGI is an intense period of reading research papers around Geometry and AI, listening to talks you may find interesting, engaging in deep academic discussions and jokes you barely understand (I mean, what are Psycho-acoustics?), learning math for those without a strong math background, acquiring coding skills for students new to programming, and using them to solve hard research problems for an entire summer, all while learning from Rock-star professors and brilliant students from around the world. (Without forgetting the generous stipend and a big swag box shipped from Boston, MA.)

My SGI Experience

July 8th was the much-anticipated day. That serene evening, we were officially welcomed by Prof. Justin, SGI chairman and organizer. My heart boggled with joy as I finally met him and other fellows — now friends — like Megan Grosse, Aniket Rajnish, Johan Azambou, Charuka Bandra, and a few others. Prof. Justin opened the floor for the tutorial week and provided a brief overview of what the upcoming weeks would entail.

The Tutorial week was a perfect blend of fun and fast-paced learning, including a power outage which lasted 3 months haha. Right after Prof. Justin's welcoming, we had our first tutor — the "Incredible" Professor Oded Stein, a Computer Science professor from the University of Southern California and tutorial week chair for SGI '24. Prof. Oded introduced us to Geometry Processing and its significance to various groups, from artists to programmers. He also taught on surfaces, meshes, their representations via triangles and faces, and how to store them using object-lists and face-lists. Additionally, we explored the different types of curvatures (normal curvature, Mean curvature, Principle curvature, Gaussian curvature, and Discrete Gaussian curvature). Next, was a session on visualizing 3D data, led by Dr. Qingnan Zhou, a Senior Research Engineer at Adobe Research.

On Day 2 of the Tutorial week we learned about parameterization and its applications in computer graphics under Richard Liu (University of Chicago), followed by technical guidance from Dale Decatur. Day 3 featured Prof. Silvia Sellán from Columbia University, who explained different methods of shape representation and their trade-offs, and ended with a talk on Neural Fields by Towaki Takikawa (University of Toronto). On Day 4, Derek Liu (from Roblox) covered mesh simplification techniques and LoD strategies, and Eris Zhang (Stanford) dove into deeper technical concepts relevant for research. The final day, led by Dr. Nicholas Sharp (NVIDIA), focused on mesh quality and remeshing, while Zachary Ferguson (MIT) discussed floating-point issues in collision detection. Unto real work!

During the Research Weeks, I undertook 4 different research projects within 5 weeks, all while also attending the Annual CMFI International Bible Camp for two weeks and hacking for 48 hours at CITS.

Project 1 - How to Match the Wiggleness of Two Shapes

Project 1 (Week 2) was led by Dr. Nicholas Sharp, senior research scientist at NVIDIA. Our research topic focused on how "Well" various surfaces can approximate deforming meshes. I learned about chamfer distances, the Gromov-Hausdorff distance, and the polyline algorithm. Source code. We concluded the week with our first group article on How to Match the Wiggleness of Two Shapes, published by Artur Bogayo.

Project 2 - Pseudo-Rendering for Cortical Mesh Segmentation

Project 2 (Week 3) was my favorite!! Dr. Karthik Gopinath, a visiting mentor from Harvard University, led us to address a key question: "How can the accuracy and efficiency of existing 2D cortical mesh segmentation techniques be improved for neuroscientific tasks?" After several failed attempts, we implemented a PyTorch-based deep learning pipeline, which we termed "pseudo-rendering". This technique converts 3D cortical surfaces into multiple 2D projections, trains convolutional neural networks on those projections with data augmentation to improve generalization, then maps the segmented 2D outputs back onto the 3D mesh through inverse projection to produce per-vertex parcellations, giving Neuroscientists around the world a new way to segment their brain surfaces, for easier disease detection and treatment. If you're interested, you could take a look at our source code. Additionally, I extended the project for a month after the SGI, to iterate and validate our method against standard segmentation baselines, then documented the full workflow and visualizations in a research article published in MIT SGI 2024.

Projects 3 & 4 - SLAM and Neural Implicit Geometry

During the next two weeks, I supported a team effort benchmarking state-of-the-art SLAM systems (DROID-SLAM, SAGE-SLAM and Gaussian-SLAM) for real-time reconstruction on clinical endoscopy data under the supervision of Prof. Roni Sengupta, Assistant Professor at the University of North Carolina at Chapel Hill. I also participated in a comparative study of neural implicit geometry representations, exploring how different architectures and loss functions affect shape reconstruction fidelity, working with Dr. Paul Zhang from BackFlipAI. These two weeks deepened my practical understanding of SLAM performance on medical data while broadening my exposure to cutting-edge work on implicit geometry and shape reconstruction.

August 16 marked the end of SGI 2024, which closed with a group photo (Don't my dark image — was traveling back from our Bible camp), with lots of virtual hugs and watery eyes. Words cannot express how grateful I am to MIT and Professor Justin Solomon. In fact, I've thanked God more than 2,500 times till date for granting me this lifetime opportunity, which I didn't deserve. The SGI taught me academic transparency and honesty toward mentors and supervisors — I learned that admitting what went wrong rather than sugarcoating is a quality people value in long-term collaborators. I also developed resilience: due to three months of power outages in my neighborhood, I spent much of my time working from public places, bars, or friends' houses — often at odd hours and exposed to uncertainty (thieves, weird stuff). That experience taught me to get things done no matter what. Above all, I discovered a deep passion for computer science research, a passion I hope to pursue further through a doctorate program soon.

APPLICATION TIPS FOR SGI 2026!

If you're an audacious college or graduate student willing to dive into the field of Geometry Processing/AI/ML, work with Top Professors and researchers, and earn some good money in a relatively short period, then consider applying for the next edition of the Summer Geometry Initiative (applications open early 2026). Below are a few tips you might find helpful!

1. Show genuine motivation and fit

SGI wants students with backgrounds in mathematics, computer science, or related fields — but they also care that your interests align with geometry processing and that you are genuinely excited about the work. Read some papers on this, or take Professor Oded's introductory course to Geometry Processing.

2. Explain your motivation clearly in your cover letter

Clearly explain why you want to join SGI — what draws you to geometry processing, how your background supports that interest, and what you hope to gain. For me, I had attended a hackathon where we used Blender to model a 3D solution to waste management in Buea, Cameroon. Yours could be anything. Just connect the dots to Geometry.

3. Highlight concrete skills and projects (not just coursework)

It helps a lot if you can show practical experience in relevant areas — coding (Python, C++, JS), math courses (linear algebra, statistics, differential equations), and ideally projects that combine math + CS + graphics. SGI explicitly allows you to include links to past work or personal webpages.

4. Demonstrate initiative via self-driven projects

Projects that integrate your math and coding skills distinguish you from applicants who only have coursework. Show that you don't wait to be taught — you go learn.

5. Be honest and authentic - share your story and background

SGI and many similar programs value diversity and non-traditional academic backgrounds. They explicitly encourage applicants from underserved backgrounds, including students from universities without strong research cultures. I clearly stated this, and expressed hopes in joining MIT for graduate studies.

6. Prepare a clean, well-organized application

SGI requires a cover letter, resume/CV, and optionally a transcript or reference letter — get to your referee early, they should KNOW you. Make sure your materials are polished: clear, concise, well-formatted; include links to your projects; and ensure your cover letter directly addresses why you want SGI and what you bring.

7. Demonstrate passion, persistence, and readiness to learn

They don't want students who just come for the money and experience, but students who actually want to pursue careers in the field. If you've maintained interest in mathematics/graphics despite challenges — limited resources, irregular power supply — mention it. This shows resilience and genuine commitment.

8. BONUS: Cover Letter

Here's the link to my SGI Cover Letter. Please, use it only as a reference to write your own story, not as some perfect template. There's no formula for success. There are many ways to Jerusalem.

I'm happy you've read so far! Be sure to apply for next year's SGI. Feel free to contact me on LinkedIn — we could chat 1-on-1 about this. Other previous Fellows like Elshadai Tegegn also shared their experiences, which you can see right here.