=20
Moravec's Paradox and the R=
obot Olympics (11 minute read)
Many of the things we struggle to progr=
am into computers can be learned from data, but only when that data is avai=
lable. If we can't learn from data, we are forced to program it in at lower=
performance. Computers can learn any skill, but only with enough data, and=
we don't want to have to need huge amounts of data for every single task. =
As models get better, they will need less data to learn new tasks, and when=
this happens, we'll finally be able to build truly general methods that co=
mbine physical understanding and cognition.
=20
|
=20
Microsoft wants to replace =
its entire C and C++ codebase, perhaps by 2030 (3 minute read)
Microsoft is hiring staff to translate =
its codebase from C and C++ to Rust. It plans to combine AI and algorithms =
to rewrite its biggest code bases. The engineer will be expected to work on=
the tools Microsoft is building to complete the project. The candidate wil=
l have to work three days a week in Microsoft's Redmond office. The role pa=
ys between $139,900 and $274,700 a year.
=20
|
|
Science & Futuristic Technology
|
=20
Single Injection Transforms=
the Immune System Into a Cancer-Killing Machine (7 minute read)
An Australian team has discovered a way=
to transform normal T cells into supersoldiers in the body. The treatment =
helped four people with stubborn blood cancer go into remission for up to f=
ive months. The trial marks a step towards the next revolution in CAR T the=
rapy. CAR T cells are usually made outside the body, and making them is cos=
tly for both the body and wallet. The durability of the new therapy, its lo=
ngevity, and immune side effects still need to be studied, but the trials h=
ighlight the promise of a one-and-done therapy for deadly blood cancers.
=20
|
=20
The ML drug discovery start=
up trying really, really hard to not cheat (Leash Bio) (33 minute read)
Leash Bio is a Utah-based startup that =
aims to build big datasets of small-molecules x protein interactions and tr=
ain a model on it, then use the trained model to design a drug. The startup=
has a nearly pathological design to make sure its models are learning the =
correct thing. The startup has produced a lot of interesting artifacts from=
its line of research. This post looks at a few of the discoveries the star=
tup has made.
=20
|
|
Programming, Design & Data Science
|
=20
Interviewing for ML/AI Engi=
neers (16 minute read)
The ML Systems Design interview, as an =
interview type, is very easy to do badly. Some companies run a parody of th=
e Big Tech interview process, while others clearly have thought deeply abou=
t what they want, but they still don't conduct effective interviews. The in=
terview process should be robust enough to avoid being fooled by people who=
are good at talking, but still accurate enough to calibrate the candidate'=
s level as well as willingness to push out-of-band offers. The best pitch f=
or a candidate is to present to them a slate of talented and thoughtful int=
erviewers who could be their future co-workers, and an interview process ri=
gorous enough to assure them that their co-workers have been thoroughly exa=
mined.
=20
|
=20
On Friday Deploys: Sometime=
s that Puppy Needs Murdering (6 minute read)
Deploy freezes are a workaround, not a =
virtue. It's possible to evade the high-stakes peril of locks and freezes a=
nd terrifying recoveries by freezing merges and letting people work on othe=
r things, and also not freezing deploys unless your goal is to test deploy =
freezes. We often forget about the outages and instability we trigger with =
our inaction. Just run the deploy process as usual, but don't ship new code=
out.
=20
|
|
=20
Steam, Steel, and Infinite =
Minds (7 minute read)
Every technological shift requires peop=
le to stop seeing the world through the past and start imagining a new one.=
We are still trying to build chatbots onto workflows designed for humans. =
What we should do instead is to imagine what knowledge work could look like=
when the busywork is delegated to minds that never sleep. The knowledge ec=
onomy is about to undergo a significant transformation.
=20
|
=20
What Jobs Are Made Of (6 mi=
nute read)
Early-career roles are traditionally mo=
re execution-heavy. Over time, contribution tends to shift toward judgment =
and agency. AI systems are making faster progress on execution than on defi=
ning problems, choosing what to work on, and navigating ambiguity. This dis=
proportionately affects entry-level hiring as the execution layer becomes c=
heaper and thinner.
=20
|
|
=20
MicroQuickJS (GitHub Repo)<=
/strong>
MicroQuickJS is a JavaScript engine tar=
geted at embedded systems that compiles and runs JavaScript programs with a=
s low as 10 kB of RAM.
=20
|
=20
Where the Horses Went (22 m=
inute read)
The core problems in robotics are far f=
rom solved, but most of the substantial blockers to robotics deployments no=
w seem solvable at a technical level.
=20
|
|
=20
| =20
Love TLDR? Tell your friends and get rewards!
|
|
Share your referral link below with friends to get free T=
LDR swag!
|
|
|
|
|
Track your referrals here.
|
=20
=20
|
Want to advertise in TLDR? =F0=9F=93=B0
If your company is interested in reaching an audience of tech exe=
cutives, decision-makers and engineers, you may want to advertise with us.
Want to work at TLDR? =F0=9F=92=BC
Apply here,=20
create your =
own role=20
or send a friend's resume to jobs@tldr.tech=20
and get $1k if we hire them! TLDR is one of Inc.'s Best Bootstrapped businesses of 2025.
If you have any comments or feedback, just respond to this email!
Thanks for reading,
Dan Ni & Stephen Flanders
|
|
|
|
![3D"Delve"](3D"https://images.tldr.tech/delve.png")