Location: Remote / Dehradun (Hybrid options available)
Engagement Model: Part-time / Contractual
Time Commitment: 8–10 Hours / Week
Role Mandate
We are soliciting applications for a Senior AI Prompt Engineering Lead to architect, govern, and optimize high-fidelity Large Language Model (LLM) systems. This role is positioned at the intersection of Agentic AI and Hiring Automation , requiring a sophisticated approach to building systems that recruit, evaluate, and interact with human talent autonomously.
This is not a content generation role; it is a systems engineering role . You will be responsible for designing the cognitive architecture of our platform, utilizing frameworks such as LangChain and LangGraph to build deterministic, scalable, and reasoning-capable agents for production environments.
Core Responsibilities
1. Advanced Prompt Architecture & Cognitive Modeling
- Strategic Design: Engineer production-grade prompt infrastructures for complex workflows, including candidate evaluation, resume parsing, interview automation, and autonomous stakeholder communication.
- Methodology Implementation: Deploy advanced prompting paradigms—including Chain-of-Thought (CoT), Tree-of-Thought, Self-Consistency, and Instruction Hierarchies—to ensure high-precision reasoning.
- Constraint Engineering: Architect robust guardrails and instruction-following protocols to maintain system safety, prevent jailbreaks, and ensure strict adherence to hiring rubrics.
2. Agentic AI & Workflow Orchestration
- System Construction: Build and manage stateful, multi-agent workflows using LangGraph and LangChain .
- Decision Logic: Design complex, multi-step decision trees that incorporate human-in-the-loop (HITL) checkpoints, autonomous error recovery, and conditional branching.
- Operational Efficiency: Optimize execution paths for latency and token cost without compromising the depth of analysis or system reliability.
3. RAG & Knowledge-Grounded Systems
- Pipeline Engineering: Architect Retrieval-Augmented Generation (RAG) pipelines that ensure high-fidelity context injection, minimizing hallucinations through rigorous source attribution.
- Vector Strategy: Manage integration with vector databases (Pinecone, Weaviate, Chroma) and implement advanced retrieval strategies such as semantic re-ranking, query expansion, and context compression.
4. Governance, Evaluation & Optimization
- Quality Assurance: Define and implement automated evaluation frameworks (LLM-as-a-Judge) to conduct regression testing on prompts and measure output drift.
- Model Selection: Make strategic decisions regarding model routing (GPT-4 vs. Claude vs. Gemini) and determine the viability of PEFT/LoRA fine-tuning versus context-window optimization.