Role Overview
The AI Engineer designs, develops, integrates, tests, and operates production-ready AI applications, with a focus on Generative AI (GenAI), Retrieval-Augmented Generation (RAG), and agentic workflows. The role turns product and business requirements into reliable software using foundation models, enterprise data, APIs, and cloud services. The AI Engineer works closely with AI Architects, software engineers, data engineers, product teams, and security stakeholders to deliver useful, secure, observable, and cost-aware AI capabilities such as conversational assistants, document intelligence, summarization, and knowledge Q&A.
Key Responsibilities
· Develop maintainable, tested Python services and components for GenAI applications, data processing, retrieval, and model integration.
· Implement prompt templates, structured outputs, tool/function calling, and model-provider integrations; evaluate and iterate based on task requirements.
· Build end-to-end RAG pipelines including document ingestion, parsing, cleaning, metadata extraction, chunking, embedding, indexing, retrieval, reranking, and context construction.
· Integrate vector and hybrid search systems; tune retrieval and grounding quality using relevant evaluation datasets and metrics.
· Develop agentic workflows with bounded tool use, state management, routing, retries, error handling, and human approval where needed.
· Integrate enterprise APIs, databases, document repositories, and business systems while respecting authentication, permissions, and data boundaries.
· Build application APIs and user interfaces or prototypes using frameworks such as FastAPI and Streamlit; collaborate with frontend engineers for production interfaces.
· Deploy and operate AI services in cloud environments; use containers, CI/CD, configuration management, and environment separation.
· Instrument applications with logs, traces, metrics, and feedback capture; monitor quality, latency, token usage, failures, and cost.
· Write unit, integration, retrieval, evaluation, and regression tests; participate in code reviews, documentation, and technical knowledge sharing.
· Apply security and responsible AI practices, including prompt-injection defenses, sensitive-data handling, output validation, and safe tool permissions.
Requirements
Mandatory Skills & Experience
- Python & Software Engineering
· Strong practical Python skills, including writing modular, readable, maintainable, and testable code.
· Experience with common Python libraries such as NumPy, pandas, and Pydantic; familiarity with typing, packaging, virtual environments, and dependency management.
· Working knowledge of REST APIs, JSON, asynchronous programming (async/await), concurrency, exception handling, and API integration.
· Familiarity with Git, code reviews, unit testing (e.g., pytest), debugging, and basic software design principles.
- Generative AI & LLM Application Development
· Hands-on experience building LLM-powered applications such as chat, summarization, extraction, document assistants, or question-answering systems.
· Understanding of tokens, context windows, prompt design, system instructions, temperature and other generation parameters, structured generation, and model limitations.
· Experience integrating hosted or open-source LLMs through APIs or model-serving endpoints.
· Ability to use streaming, caching, batching where appropriate, and asynchronous processing to improve responsiveness and manage costs.
- RAG, Embeddings & Search
· Practical experience implementing RAG: document loading, parsing, chunking, embedding generation, vector indexing, retrieval, context assembly, and grounded response generation.
· Understanding of dense retrieval, keyword search (BM25), hybrid search, metadata filtering, query rewriting, and reranking.
· Familiarity with vector/search technologies such as FAISS, pgvector, Pinecone, Qdrant, Weaviate, Chroma, Milvus, Elasticsearch/OpenSearch, or cloud equivalents.
· Ability to evaluate retrieval and answer quality using test questions, relevance/groundedness checks, and feedback.
- AI Frameworks & Agentic Workflows
· Experience with at least one current LLM application framework, such as LangChain or LangGraph; familiarity with framework abstractions and trade-offs.
· Experience or working familiarity with agentic workflow frameworks such as Microsoft Agent Framework / AutoGen, Semantic Kernel, or CrewAI.
· Understanding of tool/function calling, workflow state, routing, task decomposition, memory, retries, and human-in-the-loop patterns.
· Familiarity with MCP (Model Context Protocol) and secure integration of agents with tools or external services is desirable; where required by the role, treat as a core skill.
· Ability to implement bounded, testable workflows rather than relying on unconstrained autonomous behavior.
- Cloud & Deployment (AWS, Azure, GCP)
· Hands-on deployment experience on at least one of AWS, Microsoft Azure, or Google Cloud Platform, with willingness and ability to work across cloud environments.
· AWS familiarity: Amazon Bedrock, SageMaker, S3, Lambda, ECS/EKS, IAM, and CloudWatch.
· Azure familiarity: Azure AI Foundry, Azure OpenAI where applicable, Azure Machine Learning, Azure Functions, AKS, Entra ID, and Azure Monitor.
· GCP familiarity: Vertex AI, Gemini, Cloud Storage, Cloud Run, GKE, BigQuery, and Cloud IAM.
· Understand cloud fundamentals including compute, storage, networking basics, identity, secrets, logging, and deployment configuration.
- APIs, Data & Application Integration
· Experience integrating AI components with REST APIs, SQL/NoSQL databases, enterprise data sources, and application services.
· Understanding of data preprocessing, document formats, metadata, access permissions, and data freshness for retrieval systems.
· Familiarity with FastAPI or a similar Python API framework; Streamlit or equivalent is useful for prototypes and internal tools.
- Quality, Security & Observability
· Ability to test AI features beyond conventional software tests, including prompt/model regression, retrieval relevance, groundedness, and safety.
· Familiarity with tracing and monitoring tools such as Langfuse, LangSmith, OpenTelemetry, or cloud-native observability.
· Understanding of prompt injection, data leakage, output validation, moderation, least-privilege tool access, and secure handling of credentials.
· Ability to track operational metrics such as latency, errors, token consumption, throughput, and cost.
Good-to-Have Skills
· Model customization:
fine-tuning, LoRA/QLoRA, PEFT, quantization, prompt tuning, or working with Hugging Face Transformers.
· Experience with PyTorch or TensorFlow and foundational ML/NLP concepts, including model training or inference workflows.
· Containerization with Docker; Kubernetes or managed container services; CI/CD pipelines and infrastructure as code.
· Multimodal applications involving image, audio, speech, or vision-language models.
· Experience with model providers and model families such as OpenAI, Anthropic, Google Gemini, Meta Llama, Mistral, or Cohere.
· Frontend familiarity (JavaScript/TypeScript, React) or experience connecting AI APIs to web/mobile products.
· Experience with data processing at scale, distributed processing, or tools such as Dask or Spark.
· Exposure to AI governance and risk practices, including NIST AI RMF, ISO/IEC 42001, or regulated-domain requirements.
Qualifications & Experience
· Bachelor’s degree in Computer Science, Engineering, Data Science, or a related discipline, or equivalent practical experience.
· Typically 3+ years of software, data, or AI/ML engineering experience; practical GenAI application experience is strongly preferred. Adjust the level based on hiring needs.
· Demonstrable project or professional experience delivering software components; production deployment experience is an advantage.
· Strong problem-solving skills, curiosity, collaboration, and ability to learn evolving AI technologies.
