Please review: Exceptional Promise - SWE in ML

I’m applying for the Exceptional Promise route (~2YOE) and would love your feedback

MC

  • RL-based execution algorithms (product in company)
    • In my current firm (algo trading) I am an active contributor in R&D of innovative equities execution algorithms
    • The algorithm is SOTA in this domain (applied RL and convex optimisation)
    • I participated in research, implementation and deployment (MLOps)
    • This is a product we offer to our investor clients to better execute their trades
    • Evidence: LOS from the tech lead, repo commits, letters from investors praising our new execution
  • ML compliance automation model (product in company)
    • This was in my prevoius job at a bank
    • Using BERT model to match scripts embeddings
    • Covering compliance on bank’s loans and overdraft applications (about 10k calls a month), leading to estimated 100k annual cost savings
    • I found an innovative way to evaluate model against very imbalanced data
    • Evidence: letter of support from senior data scientist in my team, repo commits
  • Micro:bit (non-profit)
    • Contributed as a team to develop the first end-to-end ML education module on the new micro:bit V2
    • More than 1M micro:bits are distributed to Year 7 students in the UK to teach young students about digital technology
    • Potential issue: This was a proof of concept, not sure if it’s deployed for education use in the end
    • Evidence: github commits, letter of support from CTO of micro:bit
  • AI Voice-based privacy assistant (open-source project)
    • Created the world’s first AI assistant helping users with their privacy in a smart home environment, built as a skill for the Mycroft.ai voice assistant
    • Used as a research probe in Department of Computer Science at Oxford
    • Potential issue: Apart from researchers in the department my code has no other users
    • Evidence: github repo and my commits, project report screenshot, screenshot of product, workshop presentation slides, LOS from research supervisor
  • Exceptionally high salary
  • Academic achievements
    • I was accepted to study Computer Science at Oxford and graduated 5th in year (First-Class degree)
    • I received college scholarship and other university level awards for my studies

OC2:

  • Judge at OxfordHack 2022
    • This Hackathon has over 150 participants with sponsors from big tech (Google, Microsoft etc.)
    • I was on the judge panel alongside 10 others, giving out around 20 prices

OC4:

  • We are publishing the research on the execution algo in a reputable journal
  • Letter of support from professor involved in the project on my contribution

This is my first draft, would love for any general feedback. I’m also building a YouTube channel and participating in mentoring programs, which I hope can contribute to OC2.

Thanks in advance!

Hi @yumium

Number of evidences for each competency. Personally I do not recommend having 6 evidences for mandatory and then 2 for each optional! Maybe better to reduce the number of evidences for mandatory and have at least 3 for each criteria.

Thank you @Maya, do you think I could put my first MC (RL-based execution algo) as evidence for OC3 perhaps? And drop OC4 as it could take a long time before the paper gets published.

Do you have any advice on how other evidence in MC can be moved to OC2 or OC3? Thank you!

Hi @yumium you have a good MC set. I would create 2-3 evidences out of it to create OC3 and focus on quantified impact along with some external evidence to prove this impact.

Your OC2 just have 1 evidence, needs minimum 2. Try to get the letter from organizer of the event/panel to write about your contribution and appreciation. Only giving prize in a panel is not strong enough - do you have more proof of contribution?

In OC4, an ongoing unpublished paper won’t be counted.

1 Like