Support and Engagement Specialist - AI/LLM/GenAI
Budget / SalaryHourly project
TypeFreelance project
LocationRemote
Posted1 hour ago
Role Overview & Engagement
* Role Title: Project Technical Support - AI/LLM/GenAI
* Engagement & Commitment: Part-time project support requiring approximately 2 hours per day.
* Compensation: INR 350-400 per hour.
* Working Style: Independent execution backed by strong client communication and technical teaching abilities.
* Environment: Cloud and offline/restricted environments with a Copilot-enabled development setup.
Core Responsibilities
* Issue Resolution: Quickly understand client questions, isolate technical problems, locate relevant code or configurations, and provide practical resolutions.
* Debugging: Troubleshoot Python, FastAPI, LLMs, RAG, agentic workflows, model inference, data pipelines, and deployment issues by reviewing logs, traces, API responses, and test results.
* Development & Maintenance: Support AI/LLM applications, semantic search, agent workflows, CI/CD, AWS deployments, Git workflows, and technical documentation.
* Mentorship: Teach and mentor client teams on topics ranging from LLM fundamentals to advanced concepts like RAG, vector search, and agentic AI.
Required Technical Stack
* Python & Backend: Strong practical Python programming, FastAPI REST APIs, Pydantic models, async endpoints, middleware, and tools like pytest.
* LLM / GenAI / Agentic AI: LangChain, LangGraph, Google ADK, Hugging Face ecosystem, prompt engineering, and structured output/tool calling.
* RAG & Search: Ingestion pipelines, chunking strategies, embeddings, dense/hybrid retrieval, reranking, and advanced patterns like GraphRAG and agentic RAG.
* ML & Deep Learning: PyTorch, TensorFlow, transformer architecture internals, inference optimization (KV cache, quantization), and fine-tuning concepts (PEFT/LoRA/QLORA).
* AWS & Deployment: AWS Lambda, IAM, S3, CloudWatch, API Gateway, Docker containerization, and model-serving concepts.
* CI/CD & Engineering: Advanced Git workflows, GitHub Actions, clean architecture, SOLID principles, and secrets/configuration management.
* Databases: SQL, NoSQL, Apache Cassandra, RDS/Aurora, DynamoDB, and Redis caching.
Testing, Quality, & Delivery
* Quality Assurance: Execute unit, component, integration, and regression testing using pytest.
* UAT & Jira: Manage stories, tasks, and bugs in Jira; map requirements to test cases; triage defects; and track technical blockers.
Client-Facing & Problem-Solving Approach
When handling live technical questions from clients, the expected workflow is:
1. Listen to and clarify the query without rushing to a premature solution.
2. Identify whether the issue stems from code, data, model behavior, retrieval, prompts, or infrastructure.
3. Locate the corresponding source code, logs, configurations, or data paths.
4. Form and validate hypotheses using concrete evidence.
5. Explain root causes and solutions in language tailored to the audience.
6. Provide workarounds, next steps, or dependencies if a final fix cannot be deployed immediately.
* Role Title: Project Technical Support - AI/LLM/GenAI
* Engagement & Commitment: Part-time project support requiring approximately 2 hours per day.
* Compensation: INR 350-400 per hour.
* Working Style: Independent execution backed by strong client communication and technical teaching abilities.
* Environment: Cloud and offline/restricted environments with a Copilot-enabled development setup.
Core Responsibilities
* Issue Resolution: Quickly understand client questions, isolate technical problems, locate relevant code or configurations, and provide practical resolutions.
* Debugging: Troubleshoot Python, FastAPI, LLMs, RAG, agentic workflows, model inference, data pipelines, and deployment issues by reviewing logs, traces, API responses, and test results.
* Development & Maintenance: Support AI/LLM applications, semantic search, agent workflows, CI/CD, AWS deployments, Git workflows, and technical documentation.
* Mentorship: Teach and mentor client teams on topics ranging from LLM fundamentals to advanced concepts like RAG, vector search, and agentic AI.
Required Technical Stack
* Python & Backend: Strong practical Python programming, FastAPI REST APIs, Pydantic models, async endpoints, middleware, and tools like pytest.
* LLM / GenAI / Agentic AI: LangChain, LangGraph, Google ADK, Hugging Face ecosystem, prompt engineering, and structured output/tool calling.
* RAG & Search: Ingestion pipelines, chunking strategies, embeddings, dense/hybrid retrieval, reranking, and advanced patterns like GraphRAG and agentic RAG.
* ML & Deep Learning: PyTorch, TensorFlow, transformer architecture internals, inference optimization (KV cache, quantization), and fine-tuning concepts (PEFT/LoRA/QLORA).
* AWS & Deployment: AWS Lambda, IAM, S3, CloudWatch, API Gateway, Docker containerization, and model-serving concepts.
* CI/CD & Engineering: Advanced Git workflows, GitHub Actions, clean architecture, SOLID principles, and secrets/configuration management.
* Databases: SQL, NoSQL, Apache Cassandra, RDS/Aurora, DynamoDB, and Redis caching.
Testing, Quality, & Delivery
* Quality Assurance: Execute unit, component, integration, and regression testing using pytest.
* UAT & Jira: Manage stories, tasks, and bugs in Jira; map requirements to test cases; triage defects; and track technical blockers.
Client-Facing & Problem-Solving Approach
When handling live technical questions from clients, the expected workflow is:
1. Listen to and clarify the query without rushing to a premature solution.
2. Identify whether the issue stems from code, data, model behavior, retrieval, prompts, or infrastructure.
3. Locate the corresponding source code, logs, configurations, or data paths.
4. Form and validate hypotheses using concrete evidence.
5. Explain root causes and solutions in language tailored to the audience.
6. Provide workarounds, next steps, or dependencies if a final fix cannot be deployed immediately.
Apply on Freelancer →
Project sourced from Freelancer.com. Applications happen directly on the original platform — we never collect your data.