Request for Review Stage 1 Exceptional Talent

Stage 1
1

Hi all,
Please help me review my application for Exception Talent, along with any framing guidance, as an ML engineer leaning towards NLP. Thanks in advance!

MC:

  1. Offer letter from Meta. (>p90 salary)
  2. Bonus + equity increase every year with ratings for 3 years.
  3. Huggingface model contributions (1500+ downloads per month for 2 of my models combined) plus academic citations of my model and enterprise adoption (Azure and Spark NLP), and Kaggle (Wins: 16th, 38th place among 2000+ participants), with grandmaster/master adoption of my public notebooks and techniques in competitions I didn’t win frequently as well.) plus Github PR to a famous NLP library with the contributor (a world-renowned ML expert)'s direct engagement. → All dumped into 1 doc.

OC1 (Innovation):

  1. New concept introduced in Ads Targeting space now adopted by Google after Meta’s release. (Press coverage for proof of market with internal attributions explicitly stating I drove the idea)
  2. New modelling technique introduced for Whatsapp scammer detection (Press coverage for proof of market with internal attributions to me)

OC3 (Significant contributions)

  1. Ads product fix and redesign leading to increased global revenue by X million $ since 2024.
  2. detection system of sex trafficking of children in Whatsapp (technical contributions). Increased violations of 10x by an external law enforcement body

Both contain design docs (and revenue attribution for the first one) as supplementary evidence.

LORs:

  1. Founder/CEO of an AI startup. (Former skip manager at Meta)
  2. Current Tech lead in Whatsapp
  3. CEO of a former education tech startup where I was sole engineer.

I have seen your contributions in this platform and would love to hear your reviews
@raphael @pahuja @Francisca_Chiedu @Akash_Joshi

2

The MC structure needs tightening. Salary at p90 is supporting evidence, not your primary anchor - the guide is explicit that “salary or remuneration information alone is insufficient; you will have to demonstrate how you have made a significant impact in the sector beyond your day-to-day activities.” Your strongest MC material is the HuggingFace models with enterprise adoption by Azure and Spark NLP, plus the GitHub PR with direct engagement from a world-renowned ML expert. Those are two publicly verifiable threads of external recognition - lead with those and bundle salary as corroboration in the same document.

On OC1, the employee innovation bar is a granted patent. Press coverage and internal attributions stating you drove an idea are supporting evidence, but an assessor cannot independently verify an internal email or attribution. If the press coverage names you personally for the ads targeting concept, that strengthens your case significantly - if it only attributes the feature to Meta, you will need a director-level LOR explicitly naming your personal contribution.

OC3 is your cleanest criterion. The ads revenue impact and the child safety detection system are both externally verifiable outcomes. The key is making sure your design docs and architecture diagrams show your specific contribution - the guide is explicit that “evidence demonstrates your personal work, not that of the company or team of individuals.”

On LORs: your current tech lead at WhatsApp could be flagged as a direct colleague. Swap that for someone external if possible - a former manager who has since moved companies, or a senior peer at another org who can attest to your work.

3

@sindhuv

An offer letter from Meta adds to your credibility, however, salary and other remuneration alone are not sufficient. HuggingFace model contributions can work if you were an author, co‑author, or if you led or made a significant contribution to the model. Regarding your GitHub PR to a well known NLP library with direct engagement from a world renowned ML expert, does that engagement explicitly acknowledge your work and substantial contribution? That could be useful, but simple engagements may not satisfy the recognition requirement.

For OC1, aside from internal attributions stating you drove the idea, were you mentioned in any press coverage? Overall, I think you need more convincing evidence for OC1. Can you provide a granted patent with a verifiable ID on Google Patents?

For OC3, can you show your technical contributions through your GitHub account if applicable, lines of code, documentation on product designs, or architecture diagrams clearly showing ownership? You can blur sensitive data since these are likely private repos. A letter from an executive in your company describing the impact of your technical contributions on commercial outcomes like the, increased global revenue by X million dollars since 2024 would also help for internal validation.

Overall, it appears you have seven evidence sets. I suggest you work on the remaining three. Having three strong pieces across the criteria, one for each, will increase your chances.

Your LOR authors appear to be fine, is the Tech lead in Whatsapp, your direct manager? A CTO or Executive will be better.

All the best.

1 Like
4

Thanks a lot for the substantial inputs @Raphael . Much appreciated. Now answers:

  1. HuggingFace model contributions can work if you were an author

HF: I am the sole author and published the models for NLP community’s usage.

  1. Regarding your GitHub PR to a well known NLP library with direct engagement from a world renowned ML expert, does that engagement explicitly acknowledge your work and substantial contribution

Yes. He closed a second PR referring to mine as the better and broader solution to his library, along with other comments in the PR thread requesting a merge before I did. My worry is that this is quite niche (although Nils Reimers himself is a very popular authority in the NLP/ML world)

I have a similar narrative in Kaggle where I initiated a model requirement and the paper authors from Google Research, India replied to those threads leading to google’s own model publishing (All within the same discussion threads). I wonder if that should be a separate evidence or bundled together. Currently I have bundled all the influencing-ML community contributions and their academic citations together as 1 evidence of MC in “Open source Community influences”

  1. OC1 and press coverage

This is my major worry. No “Approved” patents either. Meta doesn’t report individual names to press. But the internal attributions are from our official performance letters. So I am wondering if that can offset it.

  1. OC3

Yes. I have architecture diagrams / docs and code commits to strengthen it for both evidences. My first LOR covers this (from skip manager)

  1. Tech lead from Whatsapp

No. He is not my direct manager. Senior Staff Engineers are equivalent to CTO within Meta for each product. I suppose I need to make that clearer somewhere?

  1. Overall, it appears you have seven evidence sets. I suggest you work on the remaining three.

Thanks Raphael! I didn’t want to pad it with so much unreadable docs (As I saw rejections in this forum due to content-density). Would it still make sense to add 3 more documents?