[AI/ML] Application Review Request: Exceptional Promise (Final Year PhD Student)

Stage 1
1

Hello everyone,

I am a final-year PhD student at RPTU, Germany, and I am currently doing an AI research internship at Huawei London Research Center (~11 months). I want to apply for GTV visa for Exceptional Promise via the Digital Technology route this month. My research domain is centered around 3D Generative AI and Computer Aided Design (CAD). AI + CAD has a massive commercial value for any manufacturing industry (aerospace, car, robotics). So my case + letters are centered around it.

Can you please review my case and provide feedback?

Recommendation Letters

For each recommendation letter, I have provided the theme on which they can write about me.

  • My PhD supervisor, who is also the senior director of one of the most prestigious and largest AI labs worldwide (Theme: Leadership and Research Capability).
  • A professor from Imperial College London (We collaborated during my internship at Huawei). (Theme: Innovation and Commercial Impact of my work).
  • Professor from BITS Pilani, India (He is an expert in the domain and a co-author in three of my papers) (Theme: Research and work outside my occupation) .

Mandatory Criteria: Emerging Leader Recognition

Document 1: Independent External Validation

  • Supporting Letter from a senior research scientist at Autodesk UK.
  • Supporting Letter from a CAD startup founder in the UK.
  • ML Blogs + YouTube Videos that featured my work.

Document 2: Research Overview, Citation Metrics & Publication Rankings

  • Research Overview.
  • Google Scholar profile screenshot with total citation count (~187 in 2+ years).
  • H5-index table showing the rank of the conferences globally.

Document 3: Competitive Institutional Selection

  • Proof of high salary at Huawei + Contract of my previous Research Assistant job in Germany (shows I was part of the prestigious European Horizon project).

OC2: Contribution Beyond Employment

I will mention that all the evidences are voluntary and I was not compensated for that.

Document 4: Peer Review Service

  • Reviewed 15+ papers. Reviewer invitation emails from NeurIPS, ICML, AAAI, CVPR, SIGGRAPH Asia, ICPR.
  • I was one of the technical committee members in the ICCV 2023 Workshop (Screenshot).

Document 5: Voluntary Mentoring & Teaching

  • Guest lecture on 3D Computer Vision (screenshot of the course slide).
  • Mentoring/Co-supervision of two MSc students (One published in AAAI and another got the highest score for master’s thesis).
  • One supporting letter from the course instructor, who was also the examiner of my mentored students, + screenshot of the submitted thesis page with my name as a co-supervisor. Also, my professor in his LOR mentions my involvement in mentoring.

Document 6: Open Source Contribution

While my open-source contributions are based on my research papers, releasing code and datasets is not a requirement of my PhD or internship. It’s a voluntary effort to give back to the research community. We are also not compensated for this.

  • Github. One of my repos has 360 stars and 60+ forks.
  • Huggingface. The same paper also has a public dataset that has been downloaded 18K times in < 2 years.

OC4 — Academic Contribution

  • Document 7: 1 CVPR highlight paper (First author): Abstract + screenshot (highlight) + reviewer comments about novelty, practical applications.
  • Document 8: 1 NeurIPS Spotlight highlight paper (First Equal Author): Same as above
  • Document 9: 1 CVPR paper (Second Equal Author): Same as above
  • Document 10: 1 AAAI paper (Second Equal Author): Same as above + supporting letter from a senior scientist from Autodesk USA mentioning that the field is moving towards this direction and we are the first ones to do it.

Even though I have 3 equal author papers, in his LOR, my supervisor discussed that all the ideas originated with me and how I led the other students, as he closely watched our discussions and work.

I have a few questions:

  1. Each piece of evidence is structured like this: Page 1 (cover sheet discussing what the evidence is all about), Page 2-3: (Evidence). Is this correct format?

  2. For evidence, can I add multiple images/screenshots to one page? Also, is it acceptable if the evidence starts from page 1 with a brief cover sheet?

  3. I also have one filed patent (a week ago) to the European Patent Office. It’s not granted yet, just filed. I am the lead inventor. Should I add it?

Please review my application and let me know if any section needs more work or structuring.

Thank you for your time.

2

@msk1999 Trust you are good.

For Tech Nation, you want to simply show that you are either a potential or leading talent in any of the disciplines listed in the guidance, in your case, I suppose AI or NLP Expert. So, if you are heavy on the research narrative, you may want to consider UK Research and Innovation (UKRI), The Royal Society, and similar bodies. Also, regarding the massive commercial value of your 3D Generative AI and Computer-Aided Design (CAD) work…you may need to consider the Innovation Founder route, because the TN application is mainly focused on what you have done in the last 5 years and your ability to show that you can sustain this impact in the UK.

You said 'So my case + letters are centered around it.’ - If your case is centered around only the above, it may be weak for a TN application. However, let’s have a quick review.

Recommendation Letters most of your recommenders appear to be academics. You need to confirm that they are also recognised experts in the sector, as academics are not necessarily experts in technology.

On MC, starting your evidence presentation with a letter will weaken your narrative, and reference letters on their own are supposed to be referencing an evidence. An ML blog is not convincing or acceptable evidence. YouTube can be, but it depends on the quality, depth, and reach of the videos. For Document 2, a research overview and an H5-index table showing the rank of conferences globally does not demonstrate how you have been recognised. Citations are good, but only if the papers are sector focused and published in top‑tier journals. Finally, salary can support other evidence but is not strong on its own.

For OC2, the peer review is more of an MC evidence, so I suggest you move it there. Course slides for your lecture do not show recognition or impact. Mentoring two MSc students may be dismissed as not sector‑focused nor substantial. Yes, the 18k dataset downloads are good. For GitHub, if it’s only one repo that has the metrics, it may be seen as not substantial, but it can complement the 18k‑download evidence.

For OC4, co‑publishing papers is okay, but what truly shows that you have made academic contributions through research are the citations, references, and reach of the papers. So you need to bring the MC citations here, if these papers are part of them. Alternatively, strategically list all the papers on Google Scholar and show the total citations on the page. Along with a reference or endorsement letter from reputable academics speaking about your exceptional ability in the field, then OC4 can be okay.

I think you have a chance, if presented strategically. Also, you need some work on your narrative, as I mentioned earlier.

All the best.

3

Hello @Raphael,

Thank you for your detailed review. I will try to change my narrative from CAD-specific to broader Generative AI topics and include that in my future plans.

For LOR, two of my referees also have industry positions. The professor from Imperial (citations 20K+) is a senior manager at Huawei, and my PhD supervisor (20+ YOE) also works at a research center and is a well-known figure in Germany. I will ask them to include that.

For MC, since in AI/ML domain, conferences now rank higher than journals, and I currently have rank #1, #2 and #4 papers, I wanted to position my recognition on that. After your review, I will change the narrative. Maybe I can include the reviewer comments of my papers regarding novelty and practical impact.

For OC2, can I move the reviewer invitation email to MC and keep the actual reviewing task done by me in OC2? Because one can reject the invitation as well. I will keep the dataset and the code as they are and move the mentorship + course as the third document.

For OC4, Individual paper + Citation. The senior research scientist at Autodesk USA can provide the endorsement letter. They regularly publish papers in the same domain.

4

I’m glad I could be of help.

The manager at Huawei is more appropriate, but 20K citations is more of an academic metric. I still suggest checking the LinkedIn profile of the person who works at the research centre. Being a well known figure does not necessarily mean they are an expert in the sector.

For MC, I suggest moving everything related to the paper review into that category. The invitation email is only a recognition element; you still need to demonstrate what you actually did, the real review work.

For OC4, you can include your individual papers, the ones you co‑authored, the citations, and then the endorsement letter.

All the best.

1 Like
5

Hi @Raphael,

Thanks for the update.
Just a question, if I move the peer reviews to MC (Email + Work), my OC2 is only now (1) Github, Dataset (2) Mentorship. I feel now it’s a bit weak.

6

Peer review is more of an MC evidence than OC2, in fact I can argue it that it won’t rightly meet OC2. I suggest you look for other evidence to complement OC2.

7

Your evidence are limited to your research. All you evidence won’t meet 3 criteria. Besides you recommenders are mostly academics.

Finish your PhD, then get reference letter from UK academic you have worked with and apply through the research endorsement bodies.