Applied AI Engineer

Zenopsys is an ambitious, seed-funded startup founded by BITS alumni and based in Bengaluru. We are building an AI-native platform that automates regulatory document authoring for life sciences manufacturers, turning weeks and months of manual work into hours.

Our mission is to revolutionise life sciences manufacturing operations with AI by solving critical compliance pain points for medium- to large-scale enterprises. The work we do sits at the intersection of document intelligence, regulatory compliance, scalable software systems, and applied AI — with a direct impact on helping promising healthcare reach patients faster.

We are a small, focused, and highly talented team with experience building scalable software and machine-learning systems for millions of users at companies like Swiggy and Hotstar. As an early team member, you will work closely with the founding team, write code, design systems, and build AI agents that solve real enterprise problems from day one

We're looking for an AI Engineer who wants to work on large scale problems in regulated industries. You'll build systems that parse complex scientific documents, retrieve relevant information with high precision, generate compliant regulatory content, and verify outputs against strict quality and regulatory standards.

• Document Processing: Build robust pipelines to parse PDFs (scanned and digital), Word documents, and legacy formats. Extract structure, tables, figures, and maintain document hierarchy from 600+ page source documents.
• Retrieval Systems: Design and implement hybrid search architectures combining semantic embeddings, keyword search, and metadata filtering. Optimize chunking strategies for regulatory documents where context boundaries matter.
• Agentic Content Generation: Build LLM pipelines that generate first drafts of regulatory sections with proper citation tracking back to source documents. Generate scientific and compliance documents with embedded statistics. Our largest document generation use cases span 1000s of pages!
• Verification & Evaluation: Develop constraint-based evaluation systems to verify generated content against regulatory templates and source accuracy requirements.
• Agentic Review Systems: Build conversational agents that can answer questions over internal documentation and external regulatory guidance, with accurate source attribution. Build review agents that identify lapses with internal guidelines, regulatory expectations and flag issues like how Claude review does.
• Production Infrastructure: Handle rate limiting, error recovery, caching, and cost optimization for high-throughput LLM workloads.

Core Technical Skills

• 2-4 years of hands-on experience building ML/AI systems in production
• Strong understanding of RAG architectures and their limitations
• Strong understanding of Agent architectures, context engineering and LLM performance
• You can read a research paper and decide whether the technique applies to your problem
• You're comfortable working with ambiguous requirements and figuring out what "good for the customer" looks like

Nice to Have

• Background in NLP, OCR pipelines and handling degraded document quality
• Have built agents for production preferably in a B2B environment
• Understanding of evaluation frameworks for agentic systems 

• Early-stage equity in a company solving a real problem
• Direct impact on product direction—your work ships to customers
• Competitive salary for Bangalore market
• Small team, no bureaucracy