Available usually within 2–3 weeks
Hire AI engineers who ship past the notebook.
RAG, agents, evaluation and model serving — vetted on production deployment, not Kaggle scores.
Rates
What it costs, plainly.
Fully loaded — no recruitment fee, equipment cost, benefits or bench time. Rates exclude GST for Indian clients.
| Level | Experience | Per month (INR) | Per hour (USD) |
|---|---|---|---|
| Mid-level | 3–5 years | ₹2,50,000 | $30 |
| Senior | 5–8 years | ₹3,80,000 | $45 |
| Lead | 8+ years | ₹5,00,000 | $60 |
What they do
Day to day.
- Build evaluation suites before building the feature
- Design retrieval that actually retrieves the right passage
- Deploy and monitor models, including drift detection
- Control token spend with routing and caching
- Say when the data can't support the model being asked for
Stack
What we vet them on.
- Python
- PyTorch
- LangChain / LlamaIndex
- Anthropic & OpenAI APIs
- pgvector / Pinecone
- MLflow
- vLLM
- FastAPI
Screening
How we actually test them.
Not algorithm puzzles. Real tasks from the work they'd actually do.
- 01Building an eval set for an ambiguous task and defending the metric
- 02Diagnosing a RAG system that retrieves plausible but wrong passages
- 03Explaining a project where the honest answer was 'not enough data'
- 04Estimating token cost for a described workload
You then interview every candidate yourself. Nobody joins your team without your approval.
How it works
The terms.
Monthly rolling
One month's notice either way. No annual lock-in and no severance.
You interview
We pre-vet; you make the final call. Nobody joins your team without your approval.
Replacement guarantee
If the fit is wrong, we replace them within a week at our cost.
Your process
They work in your repo, your board, your standups. No account manager in between.
IP is yours
Assignment in the contract. They commit directly to your repository.
Equipment included
Hardware, tooling and licences covered in the rate.
Domain experience
Sectors they've worked in.
The genuinely production-experienced pool is small. Plenty of CVs list LLM projects that never left a notebook. We screen for deployment, evaluation and cost control, which cuts the field considerably.
Yes — retrieval design, chunking strategy, evaluation and guardrails are core to the screening. These are the areas where most AI projects fail.
Yes. AI engineers and data scientists are complementary — scientists find what's possible, engineers make it run reliably. Tell us which gap you have.
Other roles
Often hired together.
Get profiles
Tell us what you need.
Send the role, seniority and duration. We'll come back with profiles and availability within 4 business hours.
Usually within 2–3 weeks. You interview everyone.