Role : AI Engineer – Agentic Systems & LLM Applications
About the Role :
We’re looking for a well-rounded, forward-thinking AI Engineer who can design, build, and deploy intelligent systems powered by LLMs, retrieval-augmented generation, and agentic orchestration frameworks. The ideal candidate not only knows how to build modular, reasoning-capable AI tools but can also distill ambiguous product requirements into practical, scalable AI-first solutions.
You’ll work across the full stack : orchestrating agents, integrating retrieval systems, designing for structured outputs, and deploying models in local or cloud environments. Bonus points if you’ve explored emerging multimodal or agent communication frameworks. Most importantly, you stay current with new AI research and are excited to apply it creatively in real-world settings.
Must-Have Skills
1. LLM Application Development
- Strong experience with LLM APIs or open-source models (GPT-4, Claude, LLaMA, Mistral)
- Comfortable with prompt design, structured reasoning patterns (e.G. ReAct, scratchpads), and output validation
- Built or contributed to LLM-driven apps, assistants, or internal tools
2. Agentic System Design
Experience with LangChain, CrewAI, or similar multi-agent orchestration librariesUnderstands task chaining, tool delegation, memory / state handling, and agent coordinationFamiliar with emerging design patterns like ReAct, Planner-Executor, and Reflexion3. Retrieval-Augmented Generation (RAG)
Proficient in designing RAG pipelines using vector stores (FAISS, Pinecone, Weaviate)Knows how to retrieve, chunk, and inject relevant context to ground LLM outputAble to handle unstructured, semi-structured, and structured knowledge sources4. Structured Output & Tool Use
Experience generating structured outputs using Pydantic, JSON schemas, or custom formatsFamiliar with tool calling, using LLMs to interact with APIs, calculators, databases, etc.Comfortable validating and parsing outputs for downstream reliability5. Business & Product Thinking
Can translate high-level goals into AI-first system architecturesUnderstands tradeoffs like accuracy vs interpretability or autonomy vs human-in-the-loopCollaborates well with product and design teams to scope and iterate on solutions6. Research Fluency & Emerging Technologies
Keeps up with current trends in LLM research, open-source models, and deployment toolsFamiliarity with Model Context Protocol (MCP) or similar innovations is a plusReads new papers, explores tools (Hugging Face, Papers with Code), and experiments frequentlyGood to have :
Multimodal model experience (e.G., GPT-4V, LLaVA, OCR, visual grounding tasks)Experience with local LLM deployment using tools like vLLM, Ollama, or quantized GGUF modelsExposure to fine-tuning or alignment techniques (LoRA, PEFT, RLHF)Backend knowledge with FastAPI, Flask, LangServePrototyping in Streamlit, Gradio, or notebooks for quick internal demosCloud deployment familiarity (AWS, GCP, Azure)Awareness of hallucination mitigation, evaluation, and monitoring techniquesIf Interested, please share updated resume with anuradha.dhal@sakon.com