×
Malayy
An AI runtime that sits between models and GPUs
Last verified September 10, 2026 · Updated daily
What Malayy is building
Malayy builds a runtime that sits between AI models and the GPUs underneath them. The premise on their own homepage is blunt to the point of being a provocation: 'AI is able to harness only 1-10% true efficiency from GPUs.' Everything else follows from taking that number seriously. If the claim holds even loosely, the largest capital line in the industry is mostly buying idle silicon, and the fix is not a bigger cluster but a better scheduler. The product is called Malayy K2, and it is positioned as a transparent layer rather than a framework. It 'just works on any AI stack,' with no framework change required, which is the only distribution strategy that works for infrastructure this deep in the stack. Nobody rewrites a training pipeline for a promised speedup. They will drop in a runtime. The stated axes are unusually broad for an efficiency play: speed, cost, carbon, robustness, and models, across both training and inference. That breadth reads less like feature scope and more like a claim about where the waste actually lives. Idling and bottlenecks are not one bug. They are a scheduling problem, a memory problem, and a fault problem wearing the same coat, which is roughly the union of what these two founders have spent their careers on. The founder pairing is the thesis. Jose Faleiro's dissertation on deterministic multi-core transaction processing won the 2020 SIGMOD Jim Gray Doctoral Dissertation award, and he built Anna at UC Berkeley and FASTER at Microsoft Research — two of the fastest key-value stores anyone has shipped. Nithya Sambasivan spent years at Google DeepMind on what breaks when AI systems meet the real world. One of them makes machines go fast; the other studies why fast systems still fail in production.
Why this matters
GPU utilization is the least glamorous and most expensive unsolved problem in AI infrastructure. Compute is the dominant cost line for every lab and every serious application company, and it is being spent on hardware that spends much of its life waiting — for data, for memory, for the next kernel, for a straggler. A runtime that recovers even part of that is worth more than most model-layer optimizations, and it is worth it to everyone at once rather than to one architecture. What makes the timing sharp is the drop-in framing. The market is now full of teams who have already committed to a stack and cannot afford to re-platform, but who are being squeezed hard on inference margin. A transparent layer that requires no framework change is aimed exactly at that person. Malayy is not asking anyone to switch. It is asking them to insert. Be clear-eyed about what is verified here. The efficiency figures on the site are Malayy's own claims, unbenchmarked publicly, and the round's investors are undisclosed. What is independently checkable is the pedigree, and the pedigree is genuinely rare: a Jim Gray award winner on deterministic concurrency plus a DeepMind staff researcher on real-world AI reliability is not a team that assembled around a trend.
Open roles at Malayy
3 positions we're tracking. Roles are re-checked daily and removed when filled.
GPU Systems Engineer (CUDA, kernel scheduling, memory hierarchy)
First seen yesterday
Distributed Systems Engineer (runtime, fault tolerance, straggler mitigation)
First seen yesterday
ML Systems Researcher (training and inference efficiency, benchmarking)
First seen yesterday
Know when Malayy is hiring before anyone else
A role stays uncontested for about four days. Here's the window — and where we put you in it.
From $9/month, cancel any time.
Watching Malayy
0 applicantsRole spotted & verified
1You get the alert
1You've applied
~8Hits the job boards
250+Hiring outlook
Malayy has no job board and no dedicated careers page — that URL 404s. What it has is a direct call on its homepage inviting 'Engineers and researchers who want to solve hard problems' to write to hello@malayy.com, and $10.1M that closed days ago at a two-year-old, two-founder company. The intent is explicit; the process is not yet built. Score reflects both.
Working at Malayy
Malayy is a AI company based in San Francisco, USA. For a AI company this size, the reality is opportunity to shape your role based on the company stage.
Most Malayy jobs are based in San Francisco, USA.
How to actually get hired at Malayy
Why applying the normal way doesn't work
Malayy has an ATS but the real pipeline is referrals → sourced → recruiter picks → cold apps. You can still get through, but know where you stand.
Who to contact at Malayy
What to show them
Take one open-weights model on one GPU you can actually rent and measure where the time goes — not throughput, occupancy. Profile it properly with Nsight, break the idle time down by cause (kernel launch overhead, memory stalls, host sync, poor batching), and write up what fraction each accounts for. Then fix one of them and report the delta with your methodology and your fixed variables stated. A careful 12% with a clean method beats a claimed 3x.
A cold email that works at Malayy
What Malayy screens for
Seven undisclosed investors and a two-year stealth run says this round was assembled quietly, on reputation, before there was anything public to evaluate. That is what a raise looks like when the founders do not need to explain their credibility. The candidate edge is timing and scarcity: a two-founder company that just closed eight figures, with no recruiter, no job board, and an email address on the homepage. There is no application funnel to be filtered by, which means the entire selection process is whether the thing you send is interesting. For a systems engineer, that is the best possible arrangement and it will not last past the first ten hires.
Don't send a generic CV to Malayy. Mirror the job posting's language to get past automated screening.
Don't make these mistakes
This is the wrong company to send an AI-enthusiasm email to. Jose won a dissertation award for deterministic multi-core transaction processing and built Anna and FASTER; he has read your opinion about inference being important. Do not send benchmarks without a methodology, and do not send a speedup number without saying what you held fixed — that is the fastest way to be dismissed by someone who spent a decade in systems peer review. With Nithya, the specific trap is pitching her AI ethics or fairness as a way in. She did foundational work there, at DeepMind, and she has now started a GPU runtime company. Leading with her old research area tells her you read a bio and not the product.
Mistakes that kill Malayy applications
At people, a copy-paste CV is immediately obvious. It's an instant no.
Lead with their problem, not your ambition. Malayy is focused on GPU utilization is the least glamorous and most expensive unsolved problem in AI infrastructure — show you understand that.
Don't apply and wait. The median AI application gets no response ever. One follow-up at day five roughly doubles reply rates.
Applying to Malayy? Get the contact, not the form.
The Malayy interview process
4 stages · 14 days typical · take-home: yes · modelled from similar companies
We don't yet have verified candidate reports for Malayy. What follows is the typical process for a -person AI company — treat it as a model, not confirmed detail.
Interview stages
Recruiter Screen
Phone or video · 30 min
Basic qualification and logistics
Recruiter or HR
Hiring Manager Interview
Video call · 45 min
Role fit and experience deep-dive
Hiring manager
Technical/Functional Round
Video call · 60 min
Skills assessment and problem-solving
Team members
Final Round
In-person or video · 60 min
Culture fit and cross-functional alignment
Senior leadership
Malayy take-home assignment
Malayy includes a take-home exercise in their interview process. For AI roles, this typically involves a practical problem that takes 2-4 hours. Focus on clean, working code over premature optimization. They're evaluating how you think and communicate, not just the solution.
Malayy interview timeline
At days, Malayy's process is about average than typical for AI (14 days at this size).
Interviewed at Malayy?
Tell us how it went — stages, questions, timeline. Takes 90 seconds and it's how this page stays accurate for the next person.
Submit your Malayy interview experience →Malayy jobs, frequently asked questions
How many jobs does Malayy have open?
Malayy currently has 3 open roles, last verified September 2026.
Does Malayy hire remotely?
Currently, Malayy only has in-office roles in San Francisco, USA.
What roles is Malayy hiring for?
Malayy is hiring across Engineering, Data. The most recent opening is GPU Systems Engineer (CUDA, kernel scheduling, memory hierarchy).
How do I apply for a job at Malayy?
Apply directly through the links above, or read our guide on how to actually get hired at Malayy.
Does Malayy respond to cold emails?
We're still collecting cold email data for Malayy.
Who is the hiring manager at Malayy?
At this size, hiring is usually run by the hiring manager.
How competitive is it to get hired at Malayy?
Expect 100-250 applicants in the first two weeks for AI roles at this size. Apply within 72 hours for best odds.
How many rounds is the Malayy interview?
4 stages: Recruiter Screen, Hiring Manager Interview, Technical/Functional Round, Final Round.
Is the Malayy interview hard?
It concentrates on technical depth and system design rather than abstract puzzles. The stage candidates find hardest is Technical Interview.
Does Malayy give a take-home task?
Yes, Malayy includes a take-home assignment.
How long does Malayy take to get back to you?
Around 14 days across the full process.
What should I prepare for the Malayy interview?
Study technical depth and system design. At this size (), they care about self-sufficiency over textbook knowledge.
Where is Malayy based?
Malayy is headquartered in San Francisco, USA.
Get Malayy roles before they're posted
A role stays uncontested for about four days. Here's the window — and where we put you in it.
From $9/month, cancel any time.
Watching Malayy
0 applicantsRole spotted & verified
1You get the alert
1You've applied
~8Hits the job boards
250+