AI Engineer

Job Title: AI Engineer
Location: Richardson, TX 75082 (Onsite)
Duration: 12 Months

Must Have Skills:
• Python 3.11
• LangGraph
• LangChain
• Azure OpenAI/OpenAI APIs
• FastAPI
• Vector Databases
• PostgreSQL
• Docker

Nice to Have Skills:
• Columnar performance tuning
• Network operations domain knowledge
• OAuth 2.0
• Microsoft Graph API

Detailed Job Description:
Overall 8 years' experience with 5 years in AI development. PFB the technology skills required.

Core Language & Architecture
• Python 3.11
• Advanced type hints (PEP 484), static typing discipline
• Async programming (asyncio, async/await, async generators)
• aio / (async clients)
• Pydantic v2 (BaseModel, validation, settings management)
• Structured logging & tracing patterns
• Redis (pub/sub, TTL, async clients)
• REST API design & integration patterns
• Retry/backoff strategies (Tenacity)
• Concurrency patterns (parallel tool calls, task orchestration)

AI / LLM / Agent Systems
• LangGraph (state machines, conditional edges, checkpointing)
• LangChain 0.3.x (LLClienthain, StructuredTool, retrievers, prompt templates)
• ReAct-style agent architectures
• Tool-based agent design (40 tool environments)
• Azure OpenAI / OpenAI APIs (GPT-4o, deployment mgmt, rate limits*** token budgeting)
• Prompt engineering (few-shot, structured output, JSON mode)
• PydanticOutputParser / structured LLM responses
• Guardrails / PII redaction patterns
• Memory abstractions for agents
• Langfuse (trace instrumentation, evaluation, prompt management)
• LLM fallback chains & error recovery
• RAG prompt grounding strategies
• LLM fine-tuning
• Neural Network training & tuning
• Traditional ML models (random forest, k-means clustering, linear regression, etc.)
• ClientP development and consumption

Retrieval, Search & RAG Engineering
• Vector databases (Qdrant and/or Milvus)
• HNSW indexing parameters
• Filtering strategies
• Embedding pipelines (OpenAI ada-002 or equivalent)
• Batch embedding & re-indexing workflows
• Hybrid retrieval (BM25 semantic)
• Score fusion strategies
• Cross-encoder reranking (BAAI/bge models)
• FastAPI-based inference services
• LangChain retriever abstractions
• RAG evaluation metrics:
1.Faithfulness
2. Relevance
3. NDCG
4. MRR
• Trace-level RAG evaluation (Langfuse)

Data Engineering & ETL
• Prefect 2.x / 3.x
1.Flows, tasks, futures
2.Deployments (YAML)
3.Scheduling
• ETL/ELT design o Schema evolution
1.Query optimization
• OAuth authentication
• Warehouse/schema management
• PostgreSQL 16/17
1.psycopg 3.x
2.Connection pooling
3.SQLAlchemy 2.x (ORM asyncio)
4.Alembic migrations
5.Advanced SQL
6.Multi-table JOINs
7.CTEs
8.Window functions
• Timezone conversion
• Pandas 2.x (complex multi-stage transformations)
• PyArrow / columnar formats
• Azure Blob Storage (azure-storage-blob)
• Document ingestion/parsing:
1.Docling
2.Unstructured
3.python-docx
4.python-pptx

DevOps & Platform
• Docker
• Linux fundamentals

Nice-to-Haves
• Ray (distributed execution)
• Columnar performance tuning
• Network operations domain knowledge
• NOC / alarm correlation familiarity

API & Enterprise Integrations
• OAuth 2.0 (client credentials flow, token lifecycle)
• MSAL (browser service principal flows)
• Microsoft Graph API
• SharePoint
• Outlook
• Planner
• OneDrive
• Pagination
• App permissions
• ServiceNow REST API
• Table API
• Incident/change mgmt
• Bulk operations
• Splunk SDK
• Saved searches
• Async queries
• Log analysis
• Azure AD app registrations
• IPAM / OTNA integrations (nice-to-have domain exposure)

Minimum Years of Experience:
• 7 years

Certifications Needed:
• None

Top 3 responsibilities you would expect the Subcon to shoulder and execute:
• RAG prompt grounding strategies LLM finetuning
• API Enterprise Integrations OAuth 2.0 client credentials flow, token lifecycle MSAL browser service principal flows Microsoft Graph API SharePoint Outlook Planner OneDrive
• Vector databases Qdrant andor Milvus