AI productivity platform
Montlify
One workspace holding 33+ AI tools for writing, research and chat, with the accounts, subscriptions and infrastructure needed to keep them all running.
- React
- LLM APIs
- Streaming
- Authentication
Bench · AI/ML engineers
Hire senior AI/ML engineers
Engineers who have taken an LLM feature past the demo — through evaluation, latency, cost and the failure modes that only appear with real users.
$34 – $52 / hr · from $5,400 / month
You are probably here because of one of these.
If more than one is true, the problem is usually capacity rather than skill — and that is a different conversation.
- Your AI prototype impressed the board and falls over in production.
- Nobody on the team can say whether the model got better last sprint.
- Inference cost is growing faster than usage.
- You are hiring for an ML role that has been open for five months.
Capabilities
What these engineers actually do.
Written as work, not as keywords. If your problem is not on this list, say so on the call — we will tell you honestly whether it is ours.
LLM application engineering
Retrieval pipelines, tool-using agents, structured output, prompt and context management. We build the surrounding system, which is where almost every LLM feature actually fails.
Evaluation you can trust
A graded eval set, regression gates in CI, and a number your team agrees means 'better'. Without this, every model change is a guess with a confident tone.
Fine-tuning and adaptation
LoRA and full fine-tunes, distillation to smaller models, and an honest recommendation when a prompt change would have done the job for a tenth of the cost.
Computer vision
Detection, segmentation, OCR and document understanding, taken to a latency and cost budget you set in advance.
Inference and serving
Quantisation, batching, caching, GPU scheduling and the serving path. Typically where the largest cost reduction in an AI product is hiding.
MLOps
Training and deployment pipelines, feature and model registries, drift monitoring, and rollback that works at 2am.
Stack
The tools our bench works in daily.
We work inside what you already have. Nothing on this list is a recommendation to migrate.
Models & frameworks
- PyTorch
- Hugging Face
- LangChain
- LlamaIndex
- vLLM
- scikit-learn
Providers
- Anthropic Claude
- OpenAI
- Google Vertex AI
- AWS Bedrock
- Together
- Self-hosted
Retrieval
- pgvector
- Pinecone
- Weaviate
- Qdrant
- Elasticsearch
- Hybrid + rerank
Ops
- Weights & Biases
- MLflow
- Ray
- Modal
- Kubernetes
- Triton
Rates
Published, so you can compare before you call.
Full-time, one engineer, includes ML review by our PhD lead.
| Role | Seniority | Rate |
|---|---|---|
| AI/ML Engineer | Mid-senior · 3–5 yrs | $34 – $42 / hr |
| Senior AI/ML Engineer | Senior · 5–8 yrs | $42 – $52 / hr |
| ML Research Engineer | PhD-level | $58 – $75 / hr |
| LLM Platform Engineer | Senior · 5+ yrs | $45 – $55 / hr |
Rates are per hour for full-time engagement and include our management, review and replacement guarantee. Part-time and pod pricing differ — see services.
Working to a fixed budget, or need part-time, long-term or several engineers? Those price below the card.
Contact for pricingThe 2026 rate card and a sample contract
Every role, every rate, our standard MSA and NDA, and the two-week trial terms. No call required to read it.
Week one
What happens in the first five days.
No two-week onboarding. The first week produces something you can read.
- Reads your existing pipeline and writes down where it will break first.
- Builds a small graded evaluation set from your real traffic.
- Ships one contained improvement — usually retrieval or chunking, rarely the model.
- Reports a baseline number your team can argue with.
Measured
Numbers from delivered work.
61% → 91%
Answer accuracy on a client RAG system, without changing model
4×
Inference cost reduction on a production vision workload
11 days
Median time from contract signed to first merged PR
Hiring path
How you get one of these engineers.
The same five steps regardless of discipline.
- 01Day 0
A 30-minute call
You describe the gap. We tell you which discipline it actually is — that answer changes about a third of the time — and whether we have the person.
- 02Within 5 days
Two or three profiles
Real engineers with availability, not a database dump. Each one comes with a written note on why they fit and where they would struggle.
- 03Days 5–9
You interview them
Your process, your bar, your rejection. We do not present anyone we would not hire ourselves, and the engineer who interviews is the engineer who ships.
- 04Days 10–24
Two-week paid trial
They join your standups and open real pull requests. Stop inside the trial for any reason and the engagement ends there.
- 05Ongoing
Embedded, and reviewed
Monthly rolling from there. A senior lead reviews their work independently of you, and we tell you before you have to ask.
Related work
Something we shipped with this stack.
Questions about hiring ai/ml engineers.
Can they work with our existing model provider and contracts?
Yes. We work inside whatever you have — Anthropic, OpenAI, Bedrock, Vertex, or self-hosted weights on your own GPUs. We do not resell inference and have no incentive to move you.
Do you handle data that cannot leave our environment?
Regularly. Engineers work in your cloud accounts, on your machines or ours under your policy, with signed NDAs and per-engineer access scoping. Several of our engagements run entirely inside a client VPC.
What if the honest answer is that we don't need ML?
You'll get that answer, in week one, in writing. It has happened. It costs us a month of billing and buys a client who comes back.
Need a different discipline?
Most teams start with one engineer and add a second from a neighbouring bench within two quarters.
Backend engineers
Service and API engineers who have carried a pager, and who write the boring, well-tested code that keeps you off one.
$28 – $45 / hrSee the bench
Data engineers
Pipeline and warehouse engineers who care whether the number in the dashboard is correct, and can prove it.
$30 – $47 / hrSee the bench
Cloud & DevOps engineers
Infrastructure engineers who reduce the number of things that can page you, and who treat your cloud bill as an engineering artefact.
$29 – $46 / hrSee the bench
Tell us about the ai/ml engineer role.
Thirty minutes, an engineer on the call, and two or three real profiles within five working days.