CMU PhD on the job market: choosing problems over keyword-defined research identities
simi_97k · x · 2026-11-02
In a blog post, CMU LTI PhD Simran Khanuja reflects on how she picks research problems while on the industry job market. Employers screen by keywords — pre-training/mid-training/post-training, SFT, RL, evals, alignment, etc. — but she refuses to be defined by any single label, calling herself a "problem person" with explicit criteria instead.
The post explores how the definition of "the right problem" differs between industry (product-driven) and academia, and argues that in an era where foundations are strong enough to execute ideas quickly, choosing well is becoming an increasingly important skill.
Related event: CMU PhD Candidate Simran Khanuja Enters Job Market(2 posts)→
More from AGI Musings
- Robotics is having its GPT-2 moment, says State of AI report author Nathan Benaich — nathanbenaich · 2026-11-03
- Sabbatical researcher maps 5 theses: bifurcated software, personal agents, attention repricing — manosaie · 2026-11-02
- A PhD student argues industrialized paper production is hollowing out doctoral education — tianyin_xu · 2026-11-02
- Peter Diamandis: in the superintelligence era, purpose is the new scarcity — xmercury_one · 2026-11-02
- Ethan Mollick: labs treat products as disposable until AGI makes them obsolete — emollick · 2026-11-02
- Recent hacks support the case that LLMs make experts 100x more effective, not randoms expert — mariofilhoml · 2026-11-02