AI Research Engineer

About Maple

At Maple, we are building Cognitive Infrastructure. We believe that the biggest leap in human productivity won't come from faster processors, but from better context. We are creating the first AI that understands your professional world as deeply as you do—mapping every project, relationship, and nuance into a unified cognitive layer.

About the Interpretability Team

The Interpretability team at Maple is tasked with one of the hardest problems in modern AI: Trust. We don't just want a model that generates text; we want a system that can reliably map a user's entire professional history and retrieve the most relevant context with surgical precision. We are building the tools to look inside the 'black box' and ensure our AI is both accurate and safe.

We focus heavily on Mechanistic Interpretability, Vector Search at scale, and hybrid RAG architectures. We believe that to build a tool people trust their professional life with, we must be able to explain exactly why the AI made the connections it did.

About the Role

As an AI Research Engineer, you will bridge the gap between high-level research and production-ready systems. You'll work on everything from fine-tuning open-source models to designing novel RAG (Retrieval-Augmented Generation) architectures that can handle massive, real-time datasets with sub-second latency.

Responsibilities

• Research & Implementation: Develop and deploy advanced RAG pipelines for real-time context retrieval.
• Model Interpretability: Conduct research into model reliability and explainability within our core product.
• Optimization: Fine-tune LLMs for specific tasks like summarization, task extraction, and contextual search.
• Inference Scaling: Optimize inference latency for local and edge-based AI processing.
• Product Collaboration: Collaborate with product teams to turn research breakthroughs into user-facing features.

You may be a good fit if you

• Have a strong background in Machine Learning, Mathematics, or Computer Science (PhD or significant equivalent industry experience).
• Possess deep experience with PyTorch, JAX, or similar frameworks and have deployed models to production.
• Have published research or built significant open-source projects in the LLM/NLP space.
• Possess a deep intuition for how models work and how to measure their performance beyond simple benchmarks.

Strong candidates may also have

• Experience with **Mechanistic Interpretability** tools and techniques.
• Deep knowledge of **Vector Databases** (Pinecone, Weaviate, Milvus) and retrieval strategies.
• Experience with **CUDA** programming or low-level kernel optimization.

Logistics & Compensation

• Annual Salary: ₹15L – ₹30L.
• Equity: 0.1% – 0.4% founding equity.
• Location Policy: Indore hub or Remote.
• Interview Process:
  1. Intro Call (30m): Discussing your research interest and our vision.
  2. Research Deep Dive (60m): Reviewing your previous work or a specific paper.
  3. System Design (90m): Architecting a RAG system for 100M+ documents.
  4. Founder Alignment (45m).

How we're different

We are a research lab embedded in a product company. We don't believe in research for research's sake—we believe in research that transforms how people work. We value long-term thinking over short-term hacks.

Come work with us!

Join us in building the trust layer for the future of AI. Let's make black boxes transparent together.