Manuel Martín Salvador

AI research 🤝 Software engineering

From research to production: computer vision that works with your real data.

I work remotely from Granada (Spain) for clients worldwide.

Contact me

What I do

I help companies and R&D teams design, build, and deploy computer vision & AI systems, with a focus on reliability, metrics, and cost.

If you’re dealing with…

  • A pilot that “works” but fails on real-world data and edge cases.
  • Feasibility uncertainty: data, metrics, latency, cost, privacy.
  • Needing an external review of architecture, performance, or dataset quality.

I can help you get…

  • A clear plan: what to do first, what to measure, and key risks.
  • Reproducible results (not fragile demos) and data-driven decisions.
  • Practical handover: documentation, next steps, and technical support.

Background: PhD in ML, 15+ years bridging research and industry, and hands-on CV experience (face detection, monocular 3D eye estimation, hand tracking). More about me.

How we’ll work together

A simple process to quickly reduce uncertainty and move toward the right solution.

1

Discovery

We define the goal, success metrics, available data, and constraints (latency, cost, environment, privacy).

Outcome: feasibility assessment + action plan.

2

Prototype & evaluation

We build a baseline, iterate with reproducible experiments, and validate on your data.

Outcome: clear metrics + effort, cost, and risk estimates.

3

Production & handover

We land the architecture, MLOps, and monitoring. I make the system easy to maintain and scale.

Outcome: technical deliverables + handover + targeted support.

FAQ

What kind of projects are the best fit?

Computer vision and AI systems where reliability matters: feasibility, evaluation, architecture, performance, and bridging the gap from prototype to production.

What do you need to get started?

A short description of the goal, sample data (even a small subset), how to measure success, and constraints (latency, cost, privacy). If you don't have data yet, we can define a capture/labeling plan.

How do you structure your engagements?

I work on a project basis with clear scope and deliverables. We'll define the goals, milestones, and outcomes together before starting.

How do you communicate?

I adapt to your team: async when possible (for focus), short meetings when useful. I work in CET/CEST.

What tools do you use?

I mainly work with Python, PyTorch, and MLflow for experiment tracking/monitoring. I'm not tied to a specific cloud provider, and I can adapt to your environment (AWS, GCP, Azure, or on-prem).

Can you sign an NDA?

Yes. I've worked with teams with strict confidentiality requirements.

Do you issue invoices?

Yes. I'm registered as a self-employed professional in Spain and can issue invoices.

Services

Pick the format that fits your team. If you’re unsure, we’ll clarify it in a first conversation.

Icon for Computer Vision & AI consulting

Computer Vision & AI consulting

Best for: feasibility, metrics, architecture, and performance questions.

Typical outputs: technical report, roadmap, prototype, recommendations.

Let’s talk
Icon for ML system design & evaluation

ML system design & evaluation

Best for: building or scaling ML pipelines with fewer surprises.

Focus: data, metrics, MLOps, technical debt, architecture decisions.

Get guidance
Icon for research collaboration

Research collaboration

Best for: applied research support and co-authorship.

Focus: reproducible experiments and publication-ready results.

Collaborate
Icon for technical advisory

Technical advisory

Best for: decision-making with an expert external perspective.

Focus: build vs buy, vendor assessment, strategy and risks.

Request advisory

Want to talk?

Describe your case in 3–5 sentences. I’ll tell you whether it’s a fit and what the next step looks like. No obligation.