Senior ML / LLM Consultant (Audit of AI Recommender ) Offline

We already have an ML Engineer on the project (Python, classical ML, RAG, LangChain, vector DBs, AWS.) and an MVP AI solution in place.
 

We’re looking for a short-term consultant to review, stress-test, and improve this solution.Objective of the Engagement
 

Your main goal will be to audit the current AI/ML implementation (support chatbot + recommendation engine) and provide clear, actionable recommendations to make it:
 

• More accurate and trustworthy for users
• Provide solutions for the speed optimization
• Technically robust and maintainable for the team

• Cost-efficient and scalable as usage grows

   

Key Responsibilities
1. Architecture & Code Review
 

• Review current AI/ML architecture: data flows, pipeline, model choices, and integrations with mobile backend and admin panel.
• Review codebase (Python, ML libraries, LangChain/RAG pipelines, vector DB usage, prompt design, evaluation scripts).

• Assess MLOps practices: environments, versioning, deployment, monitoring, rollback strategy.

   

2. AI Support Chat & RAG Analysis
 

• Evaluate the virtual assistant:
  • Retrieval quality (documents, embeddings, chunking strategy).
  • Prompting strategy and guardrails (hallucination risk, tone, safety).
  • Latency, cost, and failure handling for LLM calls.
• Propose improvements to:
  • Retrieval and ranking
  • Prompt templates / system messages

  • Evaluation methods (e.g., answer quality, hallucination rate).

     

3. Cost, Performance & Scalability
 

• Review current infrastructure and model choices for response time and cost (LLM provider, embeddings, vector DB).

• Suggest optimizations: caching strategy, batch calls, model selection (cheaper/faster models for certain flows), index design, etc.

   

4. Collaboration & Knowledge Transfer
 

• Work closely with the existing ML Engineer to:
  • Validate proposed changes
  • Prioritize a realistic improvement roadmap
  • Share best practices in LLM/RAG & recommender systems.

• Deliver clear documentation and a short handover so the team can continue without you.

   

Required Profile
Must-have
 

• Solid experience as a ML Engineer / ML Architect / LLM Engineer working on production systems.
• Strong hands-on skills with Python and common ML / data stack (Pandas, scikit-learn, PyTorch or similar; SQL).
• Proven experience designing and shipping LLM-based applications, specifically:
  • RAG pipelines (vector DBs, embeddings, retrieval, ranking).
  • Prompt engineering and evaluation.
  • Integrating LLMs into real products (APIs, mobile/web backends).
• Experience with recommendation systems (content-based, rules/hybrid, ranking) in real-world products.
• Good understanding of MLOps: versioning, CI/CD for models, monitoring, logging, alerting.
• Ability to perform an independent audit: challenge existing choices respectfully, explain trade-offs to non-ML stakeholders.

• Comfortable writing clear technical documentation and presenting findings.

   

Nice-to-have
 

• Background in consumer apps, travel, or edtech (recommendation, search, personalization).
• Awareness of data privacy and compliance (LGPD, GDPR or similar) in ML/AI solutions.
• Type: Freelance / consulting