Applied Compute
0 open Forward Deployed Engineer roles at Applied Compute, plus 1 closed. Curated with comp bands.
On the board since Jul 2026 · no roles open right now
Our take
Why Applied Compute
Three ex-OpenAI researchers building 'Specific Intelligence' — custom models trained on a customer's own data — where forward deployment isn't the support function, it's the entire product.
- FDE type
- True forward-deployed
- Travel
- Medium
- Comp
- Undisclosed
- Stage
- Early-stage
A good fit if you're…
- Research-leaning engineers who want fine-tuning and agent deployment as the core product
- People fluent in RL, fine-tuning, and agent training on proprietary enterprise data
- Candidates who want frontier (ex-OpenAI) pedigree at founding stage
- Those comfortable betting on illiquid, fast-marked private equity over disclosed cash comp
'Specific Intelligence,' not a model wrapper
Applied Compute's pitch is that generic frontier models leave value on the table inside each enterprise, so it trains customer-specific models and agents on a client's own data using its in-house training stack. The company is built entirely around getting those models deployed and improving in production — which is why forward-deployed engineers are the delivery mechanism, not a post-sales team bolted onto a product.
The founders and the pedigree
It was founded in May 2025 by Rhythm Garg, Linden Li, and Yash Patil, all ex-OpenAI researchers. Reporting credits Garg with work on the o1 reasoning model, Patil with agentic/Codex-adjacent systems, and Li with ML infrastructure and RL training. For a candidate, the draw is doing FDE work next to people who built frontier systems, at a company small enough that the founders are still in the code.
Forward deployment is the product
FDEs embed with enterprise customers to fine-tune models and stand up autonomous agents against proprietary data, for both internal and customer-facing tasks. Because the deployment and the offering are the same thing, the role is more research-and-build than integration-plumbing: you're deciding what to train, how to evaluate it, and how to get an agent good enough that a customer keeps paying — not wiring a fixed product into their stack.
The cap table and the marks
The trajectory has been steep. A roughly $20M pre-launch round led by Benchmark valued it near $100M; total raised reached about $80M with Sequoia and Lux joining; a $500M valuation surfaced in September 2025; and by January 2026 it was in talks for a round at a $1.3B valuation, reportedly with Kleiner Perkins leading. Named early customers include Cognition, DoorDash, and Mercor.
Comp: undisclosed, and equity-heavy by design
The company does not publish a comp band, so treat any specific number as unverified. What the cap table implies is a competitive, equity-weighted founding package priced against a fast-rising private mark — meaning most of the value is illiquid stock whose worth depends on the next few rounds, not cash. If you need a known salary or liquid equity, this seat won't give you either yet.
Who this fits
Thrives: a researcher-engineer who wants to own a customer's model outcomes end to end and is energized by frontier-lab peers and founding-stage ambiguity. Bounces: anyone who wants a defined level, a published comp band, or a mature process — at this stage the role, the ladder, and the company are all still being invented, and that's the trade you're making.
Questions candidates ask
Who founded Applied Compute?
Three ex-OpenAI researchers — Rhythm Garg, Linden Li, and Yash Patil — who launched it in May 2025, with backgrounds spanning reasoning models, agentic systems, and ML infrastructure/RL.
What's the comp?
It's undisclosed. The cap table (Benchmark, Sequoia, Lux; $1.3B-range valuation talks) points to a competitive, equity-heavy package, but any exact figure you see should be treated as unverified.
Who are the customers?
Publicly named early customers include Cognition (maker of Devin), DoorDash, and Mercor.
Is this a real FDE role or just consulting?
It's FDE-native: the whole company is built around embedding engineers to train and deploy custom models on customer data, so deployment is the core product rather than a services arm.
Analysis updated 2026-09-24
Open roles at Applied Compute
Compare Applied Compute
Others in the same tier
Free forever. Unsubscribe anytime.