Middle AI/ML Engineer

Medellín, Antioquia,Bogotá, Capital District,Bucaramanga, Santander,Cali, Valle del Cauca,Barranquilla

About project

What You'll Do:

Build & Ship ML (55%)
Design and deliver ML pipelines from experimentation to production;
Build and optimize models — supervised, unsupervised, and generative AI;
Write clean, tested, modular Python code;
Deploy and monitor models; track performance and prevent drift;
Contribute to LLM applications: RAG systems and agent workflows;
Use AI coding tools on every task to move faster and write better code.
Agentic & AI-Assisted Engineering (20%)
Use Claude Code or similar AI tools to deliver client projects;
Build with agent frameworks (Bedrock AgentCore, Strands, CrewAI, or similar);
Integrate or build MCP servers for internal and client use;
Contribute features, bug fixes, or docs to the Provectus AI toolkit.
Collaborate & Mentor (15%)
Mentor junior engineers and give actionable code review feedback;
Work closely with DevOps, Data Engineering, and Solutions Architects;
Share knowledge through docs, presentations, or internal workshops.
Learn & Innovate (10%)
Stay current with ML research, GenAI, and agentic frameworks;
Propose process improvements and reusable ML accelerators;
Participate in architectural design and trade-off discussions.

What You Need:

Machine Learning
Solid grasp of supervised/unsupervised ML: algorithms, evaluation, trade-offs;
Deep learning hands-on experience: CNNs, RNNs, Transformers — training and fine-tuning;
Depth in at least one domain: NLP, Computer Vision, Recommendation, or Time Series.
LLMs & Generative AI
Experience building LLM apps with OpenAI, Anthropic, or Hugging Face APIs;
Hands-on RAG design: chunking, embedding, retrieval, generation;
Familiarity with vector databases (OpenSearch, Pinecone, Chroma, FAISS);
Understanding of prompt engineering and LLM evaluation.
Agentic Engineering (Required)
Proficient with AI coding tools (Claude Code, Cursor, Copilot, etc.) — beyond autocomplete;
Experience building tool-using, stateful agents with an orchestration framework;
Understanding of Model Context Protocol (MCP) — consume or build MCP servers;
Can write technical specs for AI execution and review/correct AI-generated output;
Aware of agent monitoring, evaluation, and cost optimization in production.
Cloud & Infrastructure
Solid AWS: SageMaker, Lambda, S3, ECR, ECS, API Gateway;
Familiarity with Amazon Bedrock (model invocation, Knowledge Bases, Agents);
Basic awareness of Infrastructure as Code (Terraform or CloudFormation).
MLOps & Data
Production ML deployment experience;
Experiment tracking with MLflow, W&B, or similar;
CI/CD pipelines for ML; model monitoring and drift detection;
Advanced Python (async/await, OOP, packaging); strong pandas, NumPy, SQL;
Docker for containerized ML workloads.
Experience & Education
1–3 years of hands-on ML engineering experience;
At least one ML model deployed to production (or near-production);
Team-based or client-facing project experience;
Demonstrated use of AI-assisted development tools;
Education: Bachelor's/Master's in CS, Data Science, Math, or equivalent practical experience.
Key Traits
Strong problem-solver — breaks complexity into testable pieces;
Clear communicator — written docs, PRs, and explanations to non-technical stakeholders;
Fluent English (B2+);
Proactive — raises blockers early and comes with proposed solutions;
Collaborative mentor who helps without creating dependency.
Nice to Have
AWS certifications;
Kubernetes experience;
GraphRAG or custom MCP server experience
Open-source contributions or published work on agentic systems.

What We Offer:

• Competitive salary based on competencies and market rates;
• Premium AI tooling: Claude Code, Cursor, and Provectus AI toolkit;
• Mentorship from Senior ML Engineers and Tech Leads;
• Clear growth path: Mid-Level → Senior ML Engineer → Tech Lead;
• Learning budget for courses, certifications, and conferences;
• Remote-first culture; work on projects across LATAM, North America, and Europe;
• Health benefits.