Infocusp Innovations - Senior LLM Engineer

About the Role :

The individual will lead the design, development, and deployment of complex, autonomous agentic AI solutions within production environments, working across multi-agent frameworks and orchestrating large language models. The position is for scalable applied AI, with a focus on delivering robust and fair agentic systems that deliver organizational value, autonomy, and operational efficiency.

Responsibilities :

- Architect and deploy scalable multi-agent AI systems with advanced tool integration.

- Build and optimize AI pipelines using frameworks like LangGraph, ADK, and LangChain.

- Integrate agentic AI into enterprise workflows to automate complex business processes.

- Review and execute code written by AI agents to ensure correctness and maintainability.

- Review agent configurations, prompts, and tool wrappers to prevent behavioral or performance regressions.

- Audit agent-generated ML workflows to catch flaws like target leakage.

- Launch, monitor, and debug distributed training jobs and auto-repair loops.

- Design evaluation suites and end-to-end scenarios for non-deterministic agents.

- Debug agent workflows and align systems with regulatory and ethical standards.

- Ensure security, transparency, fairness, and reliability across AI systems.

- Mentor engineers and promote modern LLM practices across teams

Requirements :

- 4+ years of industry experience, including 1+ years with agentic orchestration frameworks (LangChain, LangGraph, ADK).

- Proficiency in multi-agent framework design, tool calling, API orchestration, and vector databases/RAG.

- Experience with continuous monitoring, telemetry, and observability tooling for non-deterministic AI agents.

- Strong background in LLM evaluation methodologies (code-based, LLM-as-a-Judge) and end-to-end testing.

- Strong Python skills (type hints, absl, absltest) and senior-level code/prompt reviewing capabilities.

- Practical knowledge of the ML lifecycle (data preprocessing, feature engineering, model design, JAX/TensorFlow).

- Experience launching and debugging distributed training jobs across cloud platforms (AWS, GCP, Azure).

- Knowledge of MLOps practices, data privacy, model security, and AI compliance regulations.

Preferred Skills :

- On-device ML (TFLite, Edge TPU, Gemini Nano, latency and power profiling).

- Experience with Hugging Face, Neo4j, or knowledge graphs.

- Background in autonomous decision-making systems, anomaly detection, and dynamic process optimization.

- Track record of publishing, presenting, or open-sourcing agentic AI innovations.

- Strong problem-solving aptitude, collaborative mindset, and excellent communication skills.

Location : Ahmedabad/Pune