Drafted — A real example package · drafted.careers
Your free application is waiting. Continue where you left off →
Real example package

This is what you actually get

A real example, not a mockup. The role, the fit score, and every document below are real, built by the same pipeline that builds customer packages. Sample candidate · Role live when built, July 2026 · Contact names appear only in real packages
OpenAI
Forward Deployed Engineer (FDE), San Francisco
Recommended Fit score 4.4 / 5
Live posting when built: jobs.ashbyhq.com/openai/305a4b22…
Get your own package →

Built for real, open roles that fit you.

After your free one: another package like this one is $10, or up to five for $45. No subscription required.

Document 1 of 3

Outreach: the real hiring contact

Who to reach, what to say, and when. This is the part no job board gives you.

Primary target

redacted · Global Head of Forward Deployed Engineering, OpenAI Name revealed in your package

He leads the global FDE practice this role sits in. A note that shows you already run the FDE motion, scoping, delivery, and eval-driven feedback, speaks his language directly.

Connection request

Paste into the LinkedIn connect note (277 chars, limit 300).

redacted, I'm applying for the FDE role in SF. I've spent the last three years shipping LLM systems for enterprise clients: RAG, agent workflows, and the eval harnesses that keep them honest. Owning delivery from scoping to production is the work I love. Would be glad to connect.

Follow-up message

Send once they accept, or as an InMail.

Thanks for connecting, redacted. I applied for the SF Forward Deployed Engineer opening. The shape of the role matches what I've done for the last three years at Copperline AI: embedding with client teams, scoping LLM systems, building RAG and agent workflows, and holding quality with graded evals wired into CI. Two examples with numbers behind them: I cut p95 retrieval latency on a customer-facing assistant from 900ms to under 200ms while raising accuracy 14 points on a 1,200-case eval suite, and I cut one client's monthly inference spend about 40 percent at equal quality. I'm SF based and glad to travel. Could we find fifteen minutes to talk through how I'd run a deployment for one of your strategic customers?

Alternate targets

redacted · Forward Deployed Engineering (hiring for FDE roles), OpenAI In your package

Publicly hiring FDEs across SF, NYC, and Europe. Closest to the open req itself, so a short note here can move an application out of the pile.

redacted · Forward Deployed Engineering, OpenAI In your package

A practitioner on the team. Peers reply more often than execs and can give you the real read on the work and refer you in.

Cadence

  • Apply, then connect with a note. Tip: connect with the peer alternate first, they reply more often than execs and can refer you.
  • Send the intro message once they accept (or as an InMail).
  • One gentle nudge if they accepted but went quiet.
  • Two follow-ups maximum, then stop and move on.
Document 2 of 3

The cover letter

Written from scratch for this exact posting, in the candidate's voice. Delivered as PDF plus paste-ready text.

Re: Forward Deployed Engineer (FDE), SF · OpenAI

Dear OpenAI Hiring Team,

OpenAI's Forward Deployed Engineering team is doing the version of my job I most want to be doing: taking frontier models into the hands of your most strategic customers and owning the outcome. For the last three years at Copperline AI I've run exactly this motion for enterprise clients: discovery with stakeholders, technical scoping, system design, build, and production rollout, with adoption and measured workflow impact as the bar for done.

The parts of the role about evals and field feedback are where I'd add value fastest. Every engagement I run ships with a graded eval suite and regression gates in CI. That discipline caught three model-upgrade regressions before any customer saw them, and it's how I raised answer accuracy 14 points while cutting p95 retrieval latency from 900ms to under 200ms on a customer-facing assistant. I write production Python and TypeScript daily and I've carried the pager for the systems I ship.

I'm in San Francisco, the hybrid cadence works, and heavy travel is fine. Embedding with customer teams is the part of the work I like most. I'd love to show you how I scope and land a deployment.

Sincerely,
Jordan Avery

Document 3 of 3

The tailored resume

The candidate's real experience, rewritten for this role: the summary keys to the job description, the accomplishments lead with what this team cares about.

Jordan Avery · Applied AI Engineer

San Francisco, CA (sample profile)

Summary

Applied AI engineer with eight years across backend, data, and ML systems and the last three shipping LLM products for enterprise clients. I've owned delivery end to end: scoping with stakeholders, building RAG and agent systems in Python and TypeScript, and proving quality with graded evals wired into CI. I'm at my best embedded with a customer team, turning an ambiguous problem into a production system with numbers behind it.

Key accomplishments

  • Shipped 6 production LLM applications in three years at Copperline AI: retrieval assistants, agent workflows, and document pipelines used daily by teams at insurance, legal, and logistics clients.
  • Cut p95 retrieval latency on a customer-facing RAG assistant from 900ms to under 200ms by replacing a hosted vector store with pgvector plus a reranking stage, while raising answer accuracy 14 points on a 1,200-case graded eval suite.
  • Built the company's evaluation harness (graded cases, regression gates wired into CI), now standard on every client engagement; it caught 3 model-upgrade regressions before any customer saw them.
  • Led the 3-engineer data-platform pod at Northslope Data running Kafka and Spark pipelines that moved roughly 2 billion events a day at a 99.95% delivery SLO.

Experience

Applied AI Engineer · Copperline AI (2023 to Present)

LLM-powered products for mid-market and enterprise clients.

  • Scope engagements directly with client stakeholders: business problem to LLM system design, delivery plan, and measurable acceptance criteria.
  • Design and ship RAG systems end to end: chunking and indexing, pgvector and Pinecone retrieval, rerankers, citation grounding, freshness pipelines.
  • Build agent and tool-use workflows with structured outputs, retries, and guardrails for operations teams in insurance and logistics.
  • Own evaluation: graded eval suites, regression gates in CI, live quality dashboards clients review monthly.
  • Tune cost and latency in production: model routing, caching, and prompt compression cut one client's monthly inference spend roughly 40% at equal quality.

Machine Learning Engineer · Halepoint (2022 to 2023)

Fraud models, feature store, and low-latency scoring at 3,000 requests a second.

Senior Software Engineer, Data Platform · Northslope Data (2019 to 2022)

Led a 3-engineer pod: Kafka, Spark, and Airflow pipelines on AWS.

Software Engineer · Riverbend Software (2017 to 2019)

Backend services in Python and Postgres for a logistics SaaS.

Education

BS, Computer Science, University of Colorado Boulder, 2017 · AWS Certified Solutions Architect, Associate

Every document also arrives as paste-ready text, so nothing gets locked inside a PDF.

Outreach, text version
CONTACT -- OpenAI : Forward Deployed Engineer (FDE) - SF

PRIMARY TARGET
  [revealed in your package] - Global Head of Forward Deployed
  Engineering, OpenAI

CONNECTION REQUEST (277 chars, limit 300)
---
[name], I'm applying for the FDE role in SF. I've spent the last three
years shipping LLM systems for enterprise clients: RAG, agent workflows,
and the eval harnesses that keep them honest. Owning delivery from
scoping to production is the work I love. Would be glad to connect.
---

FOLLOW-UP MESSAGE, ALTERNATE TARGETS, AND SEND CADENCE
[shown in full in your package]
Cover letter, text version
Jordan Avery
San Francisco, CA

Re: Forward Deployed Engineer (FDE) - SF, OpenAI

Dear OpenAI Hiring Team,

OpenAI's Forward Deployed Engineering team is doing the version of my
job I most want to be doing: taking frontier models into the hands of
your most strategic customers and owning the outcome. For the last three
years at Copperline AI I've run exactly this motion for enterprise
clients: discovery with stakeholders, technical scoping, system design,
build, and production rollout, with adoption and measured workflow
impact as the bar for done.

The parts of the role about evals and field feedback are where I'd add
value fastest. Every engagement I run ships with a graded eval suite and
regression gates in CI. That discipline caught three model-upgrade
regressions before any customer saw them, and it's how I raised answer
accuracy 14 points while cutting p95 retrieval latency from 900ms to
under 200ms on a customer-facing assistant. I write production Python
and TypeScript daily and I've carried the pager for the systems I ship.

I'm in San Francisco, the hybrid cadence works, and heavy travel is
fine. Embedding with customer teams is the part of the work I like most.
I'd love to show you how I scope and land a deployment.

Sincerely,
Jordan Avery
Resume, text version
Jordan Avery

Applied AI Engineer - LLM Applications, RAG, Agents, Evaluation
San Francisco, CA (sample profile)

Summary

Applied AI engineer with eight years across backend, data, and ML
systems and the last three shipping LLM products for enterprise clients.
I've owned delivery end to end: scoping with stakeholders, building RAG
and agent systems in Python and TypeScript, and proving quality with
graded evals wired into CI. I'm at my best embedded with a customer
team, turning an ambiguous problem into a production system with
numbers behind it.

Key Accomplishments

- Shipped 6 production LLM applications in three years at Copperline AI:
  retrieval assistants, agent workflows, and document pipelines used
  daily by teams at insurance, legal, and logistics clients.
- Cut p95 retrieval latency on a customer-facing RAG assistant from
  900ms to under 200ms while raising answer accuracy 14 points on a
  1,200-case graded eval suite.
- Built the company's evaluation harness, now standard on every client
  engagement; it caught 3 model-upgrade regressions before any customer
  saw them.
- Led the 3-engineer data-platform pod at Northslope Data moving roughly
  2 billion events a day at a 99.95% delivery SLO.

[full experience, education, and skills sections continue as delivered]

Get your own package

A tailored resume, a custom cover letter, and the real hiring contact — for live roles that fit you.

Get started →

After your free one: another package like this one is $10, or up to five for $45. No subscription required.