AI Engineer - Vertical LLM Agents

Position Overview:

We seek a skilled and innovative AI Engineer to design, develop, and deploy industry-specific large language model (LLM) agents. The ideal candidate will have practical experience in prompt engineering, retrieval-augmented generation (RAG), fine-tuning LLMs, integrating structured and unstructured data into knowledge bases, and evaluation.

The candidate will work on various use cases, mainly for UK-based clients.

Job Type: Full-Time
Location: Working from Mumbai office
Salary Range: Competitive, based on experience

Working Hours: 10:30 am to 7:30 pm Indian Standard Time

Days of work: Monday to Friday

Key Responsibilities:

1. Knowledge Base Development and Integration

• Define Knowledge Base Scope:
• Collaborate with domain experts to identify industry-specific requirements and tasks.
• Assess and select appropriate structured and unstructured data sources.
• Data Curation and Organization:
• Collect and preprocess data from authoritative sources (e.g., research papers, databases, manuals).
• Structure unstructured data using techniques like knowledge graphs.
• Implement data cleaning workflows to ensure high-quality input.
• Knowledge Integration:
• Embed knowledge bases into LLM workflows using tools like Pinecone, Weaviate, or Milvus.

2. LLM Fine-Tuning

• Fine-tune LLMs using frameworks such as Hugging Face Transformers or OpenAI APIs.
• Use domain-specific datasets to adapt base models to specialized industries.
• Apply transfer learning techniques to enhance model performance for niche applications.
• Monitor and improve fine-tuned models using validation metrics and feedback loops.

3. Prompt Engineering

• Design, test, and optimize prompts for industry-specific tasks.
• Implement contextual prompting strategies to enhance accuracy and relevance.
• Iterate on prompt designs based on system evaluations and user feedback.

4. Retrieval-Augmented Generation (RAG)

• Implement RAG workflows to integrate external knowledge bases with LLMs.
• Develop and optimize embedding-based retrieval systems using vector databases.
• Combine retrieved knowledge with user queries to generate accurate and context-aware responses.

5. System Integration

• Build APIs and middleware to interface between the LLM, knowledge base, and user-facing applications.
• Develop scalable and efficient query-routing mechanisms for hybrid retrieval and generation tasks.
• Ensure seamless deployment of LLM-powered applications.

6. Validation and Testing

• Evaluate model responses against domain-specific benchmarks and ground truths.
• Collaborate with domain experts to refine system outputs.
• Conduct user testing and gather feedback to improve system performance iteratively.

7. Maintenance and Updates

• Implement strategies to keep the knowledge base current with periodic updates.
• Develop monitoring tools to track system performance and identify areas for improvement.
• Address ethical, regulatory, and privacy considerations (e.g., GDPR, HIPAA compliance).

Qualifications:

Technical Skills

• Programming:
  Strong knowledge of Python and frameworks like Flask, FastAPI, or LangChain for API development.
• Data Preprocessing:
  Familiarity with preprocessing pipelines for structured and unstructured data.
• LLM Proficiency:
  Experience with LLM platforms such as OpenAI GPT, Hugging Face Transformers, or similar.
• Knowledge Base Management:
  Hands-on experience with vector databases (e.g., Pinecone, Milvus, Weaviate) and relational databases (e.g., PostgreSQL, MySQL).
• Fine-Tuning Expertise:
  Proficiency in adapting LLMs for specialized domains using domain-specific datasets.
• RAG Implementation:
  Practical experience with retrieval-augmented generation workflows.
• Prompt Engineering:
  Ability to craft and optimize prompts for complex, context-driven tasks.

Soft Skills

• Strong problem-solving skills and attention to detail.
• Ability to collaborate effectively with cross-functional teams, including domain experts.
• Excellent communication and documentation skills.

Experience

• 6+ years of experience in AI/ML roles, focusing on LLM agent development and deployment in recent years.
• Demonstrated experience in designing domain-specific AI systems.
• Hands-on experience in integrating structured/unstructured data into AI models.