Feedback on Stage 1 Application for Applied AI Scientist

Stage 1
1

Hi all, I’m collecting evidence for my stage 1 application. I would greatly appreciate any feedback on my proposed structure. Is the evidence appropriate for each criterion? It’s organized pretty simply, but is this strong enough? Thank you kindly!

Also, I may need to redact or only partially explain some of the product contributions due to NDA. is that ok? Any tips for this?

Profile: I’m an AI researcher at GitHub (Microsoft) working on AI products, and I’m applying under Exceptional Promise (technical track). I had a previous position at Microsoft Research. I hold a PhD in natural language processing (AI) obtained in the UK.

Evidence Structure:

MC

  1. Product Leadership: Technical lead and primary contributor for a shipped GitHub feature (“Feature A”) featured in the Microsoft Build 2026 keynote, and published in public GitHub blog post (my name on it). There is also a tweet from Satya Nadella’s account mentioning the feature and it was mentioned in the spring Microsoft earnings report (I have the transcript).

  2. Research Recognition: International recognition through an AACL Best Paper Award. (AACL is a top conference in AI)

  3. UK-based job with high pay: job offer paperwork shows a base salary + equity grants totaling above £100k p/a, plus an early stock grant refresher awarded within the first year.

OC3 — Significant technical contributions

  1. Feature B: Lead a research project and shipped feature B which shows in evaluations improves performance by XX%, and other metrics. I can provide documentation of the technical aspects of the feature and rigorous scientific evaluations proving efficacy.

  2. Feature C: Led evaluation-based optimization of feature C, improving performance by XX% on rigorous evals, and other metrics. I can show research documents, evaluation details, etc.

  3. Letter from my manager: confirmation of personal ownership, specialist contributions, evaluation methodology, results and ultimate deployment of both contributions.

OC4 — Academic contributions

  1. Co-first-authored EMNLP Findings research paper. EMNLP is a top conference in AI, and the paper has 300+ citations on google scholar. (my most cited paper)

  2. First-authored EMNLP main-conference paper.

  3. First-authored LREC main-conference paper.

2

The MC pack here is unusually strong for Promise. A shipped feature in the Build 2026 keynote plus a public blog post with your name on it, the Nadella tweet, and a mention in an earnings call transcript is externally verifiable at the level Tech Nation looks for at Talent, let alone Promise. Pair that with the AACL Best Paper Award and MC carries itself. The salary evidence can sit as a third item but the recognition pieces are doing the heavy lifting.

OC3 is solid but the manager letter is doing a lot of work. A single internal letter attesting to both features is thinner than two independent voices. If a partner PM, product lead, or external collaborator on either feature can write a shorter piece confirming individual ownership, the corroboration strengthens both. Rigorous evaluation documents are the right supporting artefact.

OC4 with 300+ citations on the co-first-authored EMNLP Findings paper, plus a first-author EMNLP main and LREC main, is a clean set. Independent citation trails carry OC4 more than anything else, and you have them.

On the NDA question: redacting internal metrics is fine, and the Build keynote, blog, tweet, and earnings transcript are all public artefacts anyway. Explain the redaction context briefly in the personal statement and the assessor will follow it. Nothing to worry about there.

3

@Akash_Joshi thank you for the feedback. This is very helpful! The advice for OC3 makes good sense, and I will try to get a second reference from my org to support feature C.

I also appreciate your perspective given on my MC and the positioning between promise and talent. I have around 2.5 years of work experience post-PhD, which is why I aimed for promise. Do you think it would be a good idea to aim for talent instead?

4

Applying for Promise with this profile is safe. Tech Nation automatically upgrades Promise applicants to Talent when the evidence warrants, so you’re not locked out of the Talent endorsement by choosing the Promise route.

The question is whether the current pack demonstrates established leadership rather than potential. Your MC pieces read Talent-shaped in isolation. What Talent typically requires on top is several years of shipped impact and recognition, and a single strong 12-month window can still read as Promise even when the individual pieces are strong.

2.5 years post-PhD is short for a pure Talent application. Promise remains the cleaner submit here. If the assessor reads the evidence as Talent-grade, they’ll move you up on their own.

5

@nick,

While being employed by a recognised organisation can give an applicant some leverage, the first and most important point of assessment is eligibility. AI Researcher is not listed as a qualifying role, so you need to position yourself around Artificial Intelligence or Natural Language expert if your work falls within these areas. Having both industry and academic experience is a plus, and with a PhD, you won’t struggle to demonstrate how you have contributed to the sector through research in OC4, becasue that achievement speaks for itself.

On your MC, you described your contribution to Microsoft Build, where it was posted, and who spoke about it. These do not show that you technically led a team to develop it. You need to demonstrate that you directed the work of other engineers or contributors for example, artefacts showing key technical or strategic decisions you made, such as choosing frameworks, defining APIs, Slack and email correspondences, internal documentation, commit history, or project retrospectives that highlight your leadership role. After establishing this, you can then use the Microsoft Build keynote, GitHub blog post, and Satya Nadella’s tweet as external validation of your leadership. Although MC and OC2 requires recognition, recognition can be used anywhere to validate that you actually did what you claim. For example, a Best Paper Award is more suitable for OC4 because it validates that you wrote a paper capable of advancing the sector through research and were recognised for it. Also award evidence is listed under OC4 as evidence of awards received for outstanding applied work, so it should go there. A UK‑based job with high pay can complement other evidence, but it is not sufficient on its own, your overall application and narrative should show that you have impacted the sector outside your paid job.

For OC3, it is about technical, commercial, or entrepreneurial contribution to a product‑led company. As an AI researcher, it may not be immediately clear how you contributed technically, commercially, or entrepreneurially. This is why you need to reposition yourself as an Artificial Intelligence or Natural Language expert as AI researchers on the field are within this domains, so you can show evidence of owning the research direction for technical work, for example, defining the model architecture, training pipeline, and evaluation framework. You can also show scientific validation you carried out. For Feature C, your leadership in evaluation‑based optimisation is good, and presenting research documents and evaluation details strengthens the fact that you are both an industry and academic expert. A letter from a manager can be acceptable if he is not your direct line manager; otherwise, it may be dismissed as a letter from a colleague.

For OC4, your paper with 300+ citations is strong, especially in Empirical Methods in Natural Language Processing, which is one of the top tier international conferences in AI and NLP research. You can also support this with your EMNLP main conference paper and your first authored LREC main conference paper, along with the award I suggested from MC.

Overall, you have a good evidence collection, and your narrative can be strong if positioned well. Work a bit more on your MC by adding externally validated evidence, and refine the narrative and evidence presentation for OC3. OC4, with award looks good.

You are heading in the right direction.

All the best.

6

Quick clarifier on positioning. AI research at GitHub/Microsoft on NLP products sits squarely inside Tech Nation’s technical scope. You don’t need to reposition yourself as an AI or NLP expert as a separate qualifying identity. Your role and evidence already sit inside the qualifying tech remit.

The AACL Best Paper Award belongs at MC, not OC4. The Guide’s MC examples explicitly include international awards recognising exceptional contribution to the digital technology sector, and a Best Paper Award at a top-tier AI conference is exactly that. OC4 is where your citation-backed papers live, and moving the award off MC would weaken MC unnecessarily.

Your original structure holds. Chasing internal Slack messages, commit history, and API design docs as MC evidence overweights internal artefacts. The Build keynote plus public blog plus Nadella tweet plus earnings transcript are already the externally validated proof of leadership that MC turns on. Keep the pack as you have it.

7

@Akash_Joshi I respectfully disagree with your point regarding the research-related award, and I’d like to explain why for the benefit of others reading this. Every criterion has its own context, and a strong application depends on how well an applicant presents evidence that aligns with each one.

The MC award focuses on demonstrating that an applicant has been recognised as or has shown potential to be a leading talent in the digital technology sector within the past five years through specialised work. For instance, he built a pipeline, and evaluation framework that won an award. An award can serve as validation that the applicant made a notable contribution to the tech sector and was recognised for it.

Meanwhile, OC4 specifically outlines the types of contributions to the sector and explicitly lists awards for applied work. “How do I demonstrate that I have exceptional ability in the field by making academic contributions through research?” He made research contribution and the award will validate that it was valuable and recognised for it. So the award supported by his academic achievement PhD. rightly aligns with OC4 than MC.

Regarding the AI researcher role, it isn’t listed as either a technical or non‑technical discipline. There is also a dedicated research pathway, and one would expect an applicant in that area to follow that route. So the strategic approach is to identify a closely related listed discipline. For example, a Product Owner isn’t eligible for the Tech Nation application, but because their work overlaps with Project Managers, though in different environments, it’s advisable for them to position themselves as a Project Manager rather than assuming their evidence alone will prove eligibility - It’s all about presenting ones narrative, document in a less confusing way, to reduce doubt, make assessment time faster and to increase chances.

image
image1743×1401 340 KB

8

The Guide gives you flexibility on where the AACL award sits. Both MC and OC4 are defensible homes for a top-tier conference Best Paper Award, since MC lists international awards recognising contribution to the tech sector and OC4 lists awards for applied work.

For your specific pack the decision comes down to which criterion needs the lift. Your MC without the award is the shipped feature plus the salary piece, which sits at the minimum of two items and leans on one anchor. OC4 without the award is three first-author papers with 300+ citations on the top one, which is already a strong three-piece pack that doesn’t need reinforcement.

Keeping the award in MC gives you three genuinely strong MC items instead of two, and OC4 still stands on the papers alone. That’s the practical trade. Swapping it to OC4 tightens an already-strong criterion at the cost of thinning MC to the minimum.

9

Well I get your point. But the guidance is clear that you need to meet one MC and two OCs. So, it’s strategic to put evidence that rightly align to each criterion where they belong to increase the chances of hitting the minimum requirements.

10

Nick, to close the loop for your submission: your original structure holds up. MC with the shipped feature, the AACL Best Paper Award, and the salary evidence gives you three genuinely strong items that each stand on external verification. OC3 with the two features plus a corroborating letter from a partner PM or product lead alongside your manager. OC4 with the three papers.

The award placement is a judgment call the Guide allows either way. Given your pack, MC benefits more from carrying it than OC4 does, since your papers already have the citation trail doing the heavy lifting on OC4.

11

Thank you both so much, @Akash_Joshi and @Raphael for taking the time to discuss and explain the rules in such nuance. I really appreciate the thoroughness and feel more reassured after reading your thoughts. It does look like a tradeoff, but given the balancing of MC and OC4 I’m inclined to put the paper award in my MC.

There is also a slight wording in OC4 which makes me more inclined toward MC. OC4’s language for an award:

  • Evidence of awards received for outstanding applied work, supported by excellent academic achievement (a first-class degree or distinction);

I think the word “applied” here might require some argument that my research paper has application e.g. to a product area. It is a theoretical paper with empirical results on dataset evals, and AACL features this kind of work mostly, so to me it is an easier fit to the MC award language, which is more broad:

  • You have received nationally or internationally recognised prizes or awards for excellence specifically in the digital technology sector, as evidenced by the award itself, reference letter(s) from leading industry expert(s) describing your achievement, or as evidenced by news clippings or similar evidence.
12

@nick Well! I won’t over flog this. I was an academic myself. Academic is not specifically in the digital technology sector from what you quoted.

Sector award is different from academic award. Tech Nation listed award for MC and OC4 for a reason. Take a minute to think about the reason.

Can a lecturer say I am working specifically in the digital technology sector?

  • You have received nationally or internationally recognised prizes or awards

for excellence specifically in the digital technology sector.

Tech Nation was specific where the award should be in.

I gave instance of the kind of award that can rightly work in MC in your case. Finally, one reason why you want to use evidence that rightly align to each criterion is because of an appeal. Just incase you need to appeal. It will help your argument to show how your evidence rightly align.

All the best.