Versey
Senior ML / Backend Engineer
How your CV stacks up
Upload your CV to see how well it fits this job role
?%
Set Up Personalized Fine-Tuning on Versey
Company: Versey Type: Engineering — Fractional, ~3–4 days Rate: $70–95/hr (senior ML/backend contractor, London/remote market) Location: Remote (London-friendly hours) Stack: fireworks, next.js, react, braintrust, typescript Status: Open
Wire up per-user fine-tuning on Versey: users drop in reference files, we fine-tune a small model on Fireworks so the app can write in their voice. Evals via Braintrust. The hard part is keeping cost and performance observable as concurrent fine-tunes scale.
Stakeholder: Will Taylor, CEO
Minimum earning per ticket: $1.5k
Estimated hours: ~3–4 days
What you'd build
Per-user fine-tuning on Versey. A user uploads a set of reference files (their writing samples), we kick off a fine-tune of a small model on Fireworks, and the app then generates in that user's voice. The Next.js/React product side already exists; Braintrust is set up for evals. You'd own the fine-tuning pipeline end to end — file ingestion → Fireworks job → eval → routing the user to their model.
Reasons to use Rodeo
Start with a chat, not a search bar
Grad scheme, placement, apprenticeship? Not sure what you want yet — that's fine. Your agent talks it through with you and turns "I have no idea" into a shortlist.
It searches the market for you
Every day your agent scans the market matching roles against what actually matters to you, not just keywords on a CV.
Only hits
No noise. No "maybe this fits." Just roles with a clear explanation of why they're right — and where to focus when applying.
The hard part
The hard part: making cost and performance observable in a way that scales when many users fine-tune at once. Per-user fine-tunes are cheap individually and ugly in aggregate — concurrent jobs, queueing, eval cost, inference cost per user, model cold-starts. We need to know what each user costs and how their model performs, without hand-rolling a dashboard for every metric.
What's in it for you
- Self-managed work — you own the architecture decisions, not someone else's ticket
- Fireworks + Braintrust + Next.js, no legacy to fight
- Successful trial → ongoing fractional work with Versey
What we're looking for
- Shipped production fine-tuning pipelines before (any provider — Fireworks experience is a plus, not required)
- Comfortable instrumenting cost and latency in a multi-tenant setup
- Has used Braintrust or an equivalent (Langfuse, Phoenix, Arize) for LLM evals
- TypeScript / Next.js fluent
Get help with your application
Your very own career expert that helps elevate your application to the next level.
First ticket
Dive in and set up the pipeline end to end: a user can drop reference files, a fine-tune runs on Fireworks, evals run in Braintrust, and the user can generate against their model in the app. Cost and performance metrics are queryable per user. Sketches in Figma and the Versey repo are waiting for you on day one.
How to apply
All roles on Call My Agent use the universal call-my-agent-apply skill, run from inside Claude Code. Standard install:
mkdir -p ~/.claude/skills/call-my-agent-apply
curl -fsSL https://raw.githubusercontent.com/workflow-design/call-my-agent/main/skills/call-my-agent-apply/SKILL.md \
-o ~/.claude/skills/call-my-agent-apply/SKILL.md
Then In Claude Code:
Apply to versey-versey-personalized-fine-tuning on Call My Agent
“It took my CV and asked me questions relevant to understanding what kind of jobs to suggest for me. Suggestions were almost perfect. Jobs were exactly what I’ve been looking for.”
Jessica, London
Skills
Location