Hadas Kotek » Blog »

End of year post for 2025

Dec 31, 2025

Another year in the bag! Once again, I haven’t written as many blog posts as I was hoping for, but once again, it’s been an incredibly busy year. This year had more personal stuff than work stuff, but there was a ton of both.

Travel

Last year’s post started with work, but this year what stands out most right now is travels. It’s probably just a recency effect, and yet. In January, I attended the Linguistic Society of America’s Annual Meeting in Philly, where we held the inaugural meeting of LEXING (more below). Also in January, I made a quick trip to New York City, which I hadn’t been to since before COVID(!),1 and met up with a few friends. In February, I had a work trip to Pittsburgh, and then the following week, early March, I had a work trip to Apple Park. In April I didn’t travel anywhere (gasp!).

In May, I traveled to Swarthmore as an honor thesis examiner.2 Also in May, my father came to visit and stayed with me for about 3 weeks(!!). In June, we flew out to Yellowstone National Park and it was so pretty, one of the most interesting places I’ve ever visited. In July I spent two weeks in Vienna, one week hanging out with my brother and one week at ACL. In August, I was back in Apple Park for a few days. In September, I hosted an onsite in Cambridge, so I got to do Boston touristy things without having to travel.

In October-November I spent four weeks traveling: I spent three weeks in Asia, in South Korea, Vietnam, and Hong Kong (with short trip to Macau) and then a week in Barcelona for an office visit (with 46 hours in Boston between the two flights). And then I was very tired! I didn’t go anywhere for Thanksgiving or Christmas breaks, so no travel in December. I’ll be traveling to New orleans in early January for the LSA, and although I don’t have other trips planned beyond that right now, I wouldn’t be surprised if 2026 was another travelful year.3

Phew, I’m tired just writing this. If you follow me on Insta, you’ll find lots of pics from all the trips.

Academia

I continue to maintain my affiliation with MIT Department of Linguistics. I didn’t teach a seminar this year—I’m too busy in the spring and we missed the critical time for advertising in the fall, but I’ll probably try again next year and actually try to remember to get on the books in time to advertise properly.

I attended several conferences and workshops in 2025. I already mentioned LEXING (Linguists in Industry, Non-profits, and Government), which “aims to create a space for linguists currently employed outside academia and those who are considering such careers to come together, with the goal of fostering a community of career linguists––embracing newcomers and easing career transition points”. It was So Good! I was so happy with the submissions and conversations and the new connections I was able to make. The fishbowl format, especially, was such a fun revelation. We’re gearing up to run the second iteration in this year’s LSA–now just over a week away–and I can’t wait.

I presented internally in a couple of Apple workshops, both of which involved travel but were fun and worth it. On the other hand, I went to ACL as a visitor only and was very disappointed by the quality of the content in the main program,4 though the workshops were great. I gave a talk on AI ethics locally at MIT to MBAn students, and I was a keynote speaker at INLG, which is how I ended up with that 3-week trip to Asia because going for just one week for a conference just wasn’t going to be enough.

The only new paper with my name on it in 2025 is the Apple Intelligence Foundation Language Models: Tech Report 2025, but I completed two other manuscripts: Language and technology, an invited contribution to the Routledge Handbook of Linguistics (2nd Edition), and a soon-to-be-submitted paper on (mis)gendering in LLMs.

Mentoring and service

As always, I continue to engage in career activism (you, too, can get a job outside academia and be happy, successful, and very fulfilled!). I gave the 6th iteration of my semester-long “careers workshop” at MIT, and we’ll again go into the spring with a focus on interviews. I gave career talks at the University of Essex, University of Kansas, African NLP talk (AfriCompLing) and YNLG. And I count 20 informational interviews in my calendar this year, sharply down from last year but still at a pace of one every other week or so. I also continue to meet with two mentees on a regular basis.

I am (still) the chair of the Pop-Up Mentoring COGEL subcommittee, where we hosted 10 events in 2025. Thankfully, I am stepping down as senior co-chair of COGEL in 2026, which I am excited about.5 And I’m going into my final year as member of the Executive Committee of the LSA, and continue to serve on the Committee on Institutes and Fellowships. Next year I’ll also be a member of the Committee on Committees!

Work

Work goes steady on. The word of the year has been scaling. My team has gone from 2 to 4 members with 2 more planned to reorg into my team soon and 3 open headcount for immediate filling (one offer already out), so by March or so I expect the team to be at 9 members. At that point, I probably need to start thinking about having subteams and at least one manager, but we shall see. I am also hiring an intern for the summer and a resident for a full year starting spring 2026.6 That is So Much Hiring. I also spent a bunch of time making new questions and scoring guides for my interviews, totally worth it but so much work.

Officially, my team is now called the RAI Human Evaluation team. My team is responsible for evaluating both the Apple Foundation Models and customer-facing features across languages and modalities. This coming year we’re on track to support about 3x as many features as we did in the previous two years. I also own the foundation model policy, as well as the safety taxonomy for the RAI team. It’s A Lot. (Did I mention Scaling?) So what does that mean? Recently, at least, I’ve been thinking about dataset construction, data quality, LLM-as-judge (autograding), rubrics, streamlining metrics and reporting, and the meaning of life.

Our broader org has gone through some major changes, with AIML being broken apart. My team is now officially a part of SWE. But this change was so recent that at this point it’s hard to know what it really means for us. I am concerned about my ability to continue my research activities and travel for conferences, and about whether the general support for RAI will change, but at least for now the party line is “everything continues as usual”, so I’ll just have to wait and see.

All those things I wrote last year about long hours that are mostly PST-aligned and way way too much work at an unsustainable pace is pretty much all still unchanged. That and academic stuff and mentoring and travel means I’ve gotten close to burnout this year, but I’ve also taken time off multiple times this year and traveled for fun and I think generally did better at having a life. As my team grows and responsibilities expand, I should actively think about how to manage my life and keep my sanity; but I want to do all the things I’m doing (and more!) and I’m not yet at a place where I want to choose what not to do among all the things I want to be doing. Maybe just more time off that’s not for travel.

Personal

My health has continued to be good, thankfully. I had one minor bout of the cold in 2025, but I didn’t even lose my voice. The ICL surgery I had two years ago continues to be a great improvement in my quality of life, and as of 2 days ago I again have no cavities, hurray.

My blender continues to be a high-use item for smoothies, and my combination oven/air fryer is still a favorite. As with last year, I abandoned cooking most of the year even though I really do enjoy it, but I try to order healthy food. I also continue to have home cleaning people come by every ~3 weeks and it’s sooo nice. And my cats continue to be the best little buds.

On the home improvement front, we had to have a major renovation on our back porches, which had some significant rot. In the apartment, I hung a bunch of new art that I picked up in Asia, but I didn’t do much else. No news is good news.

Citizenship

Possibly burying the lede, in February 2024, I became a citizen of the United States.7 This extra peace of mind has enabled me to say and do things that I’d been actively avoiding previously while on a green card and/or visa, such as going to protests and speaking my (political) mind. I’ve already called my senators to Let Them Know what I think a few times.

Fun

I continue to try to go to one cultural event per month – a movie, show, musical, etc. This is what I can remember/find in my calendar for the year:

Not quite one per month but given all the travel, it’s really not bad. I also read so much this year. A total of 164 books (cf 74 last year), with several very enjoyable series. Some favorites:

Did I lose sleep to binge-reading? Yes. Do I regret it? Not enough to not keep doing it!

And… that’s a wrap!

 

Notes

  1. Ok, I was there for the LSA a year and change ago, but does that really count? ↩

  2. Yes, Swat flies in multiple examiners for each undergraduate honors student. Which, wow. But sure, I’ll take your money. ↩

  3. For example, I haven’t been to Israel since December 2022. I should probably go some time this year, hopefully (ahem) between wars. ↩

  4. It was a lot of “We threw This Data at These Models and got These Numbers”, without a bigger picture or broader implications. I think I saw one non-LLM presentation. Very little theory motivating anything. And given the numbers were probably already stale by the time they were presented–and certainly are now–it was just.. meh. ↩

  5. I’m especially excited to see what our new senior chair, Kate Lindsay, has planned for 2026! ↩

  6. Think of the Apple Residency as a one-year industry postdoc. ↩

  7. It was only after the elections, though I hope those weren’t the last ones. And I already have two other citizenships, and the green card gave me reasonable stability. But citizenship gives me an extra peace of mind, especially in today’s political climate. ↩