AI Developer at Dieture in Bangalore Urban

We are looking for an AI Developer with 2–4 years of software/AI development experience to build and evolve customer-facing AI capabilities.

This is a hands-on engineering role focused on building production AI systems that understand user conversations, maintain relevant context, retrieve information, interact with application services and tools, and generate useful, safe, and reliable responses.

We are looking for someone who understands that production AI engineering goes beyond prompt writing. You should be comfortable working across LLMs, Python, APIs, RAG, memory, structured outputs, evaluation, observability, and production debugging.

Key Responsibilities

• Develop conversational and Generative AI applications using modern LLMs.
• Build and maintain AI backend services using Python and FastAPI or similar frameworks.
• Implement prompt and context engineering, structured outputs, streaming, and tool/function calling.
• Build RAG pipelines using embeddings, vector search, metadata filtering, reranking, and contextual retrieval.
• Design short-term and long-term conversational memory and personalization mechanisms.
• Integrate AI capabilities with internal APIs, databases, and business systems.
• Develop AI workflows and agentic capabilities where appropriate.
• Implement model routing, retry, timeout, fallback, and error-handling strategies.
• Build safeguards for prompt injection, inappropriate tool usage, PII handling, and unreliable model responses.
• Develop automated evaluations and regression tests for AI behaviour.
• Monitor AI systems for response quality, latency, errors, token consumption, and cost.
• Debug and continuously improve AI behaviour based on production observations.

Required Skills

• 2–4 years of software/AI engineering experience
• Strong Python programming skills
• Experience with FastAPI or similar backend frameworks
• Hands-on experience integrating LLM APIs
• Good understanding of prompt and context engineering
• Experience with structured LLM outputs and/or function calling
• Understanding of RAG architecture
• Understanding of embeddings and vector search
• Experience integrating REST APIs and third-party services
• Understanding of asynchronous programming
• Experience with Git and standard software development practices
• Ability to write maintainable and testable production code
• Strong debugging and problem-solving skills

**** Expected AI Knowledge****

Candidates should have a practical understanding of:

• LLMs and transformer fundamentals
• Tokens and context windows
• Embeddings and semantic similarity
• RAG
• Prompt and context engineering
• Tool/function calling
• Structured generation
• Hallucination and grounding
• Prompt injection
• Model latency, quality, and cost trade-offs
• Evaluation of non-deterministic AI systems

Good to Have

Experience with some of the following would be valuable:

• OpenRouter
• OpenAI / Anthropic / Google Gemini APIs
• Mem0
• LangGraph
• LangChain
• Langfuse or LangSmith
• Redis
• MongoDB / PostgreSQL
• Vector databases
• Docker
• Kubernetes
• AWS / Azure / GCP
• Background workers and queues
• WebSockets and streaming responses

Experience building production customer-facing conversational AI applications is particularly valuable.

AI Safety & Reliability

The developer will contribute to building AI systems with appropriate safeguards, including:

• Input and output validation
• Prompt-injection protection
• Tool permission boundaries
• PII-aware data handling
• Domain-specific guardrails
• Safe fallback mechanisms
• Human escalation where appropriate
• Monitoring and auditability

What We Are Looking For

We are looking for an AI application engineer rather than primarily an ML researcher.

The role is focused on combining foundation models with software, APIs, application data, retrieval, memory, workflows, tools, evaluation, and guardrails to create reliable production AI capabilities.

A strong candidate should be able to take a requirement such as:

"The AI should understand the customer's request, identify the appropriate context, retrieve relevant information, call application services when required, and generate an accurate and safe response."

…and convert it into a well-designed, testable, production-ready implementation.

We value engineers who enjoy building, experimenting, measuring, debugging, and continuously improving real-world AI systems.