AI & RAG Engineer — Knowledge Systems
You can still review the qualifications and submit a general expression of interest for future opportunities.
Build high-precision Retrieval-Augmented Generation (RAG) pipelines and conversational AI systems grounded in authentic multi-lingual Indic texts.
01About SattvaLok
SattvaLok is a mission-driven technology platform dedicated to bridging the depth of timeless Sanatan wisdom, Indian Knowledge Systems (IKS), and cultural heritage with modern digital architecture. Unlike conventional content portals or superficial aggregators, SattvaLok builds high-integrity knowledge engines, research-backed digital archives, and responsible AI-powered discovery platforms. We adhere to rigorous standards of textual authenticity, epigraphical accuracy, and cultural nuance. Our interdisciplinary team unites software engineers, AI/RAG researchers, Vedic scholars, Sanskritists, and creative storytellers to create digital experiences that are trustworthy, inspiring, and globally accessible.
02About the Opportunity
At SattvaLok, we are pioneering the application of generative AI to sacred, multi-lingual, and deeply nuanced Indic knowledge repositories. As an AI & RAG Engineer, your mission is to eliminate hallucinations and build retrieval-augmented generation architectures that deliver mathematically grounded, citation-verified responses from classical Sanskrit, Hindi, and English corpuses. You will engineer vector retrieval pipelines, hybrid lexical-dense search systems, knowledge graph embeddings, and evaluation benchmarks that ensure every AI-generated response cites precise chapter, verse, and scholarly commentaries.
03Responsibilities & Scope of Contribution
- →Design and deploy scalable Retrieval-Augmented Generation (RAG) systems over multi-million-token Indic and Sanskrit text corpuses.
- →Implement hybrid retrieval architectures combining dense vector search (Qdrant/Milvus/Pinecone) with BM25 lexical search and entity knowledge graphs.
- →Develop chunking, embedding, and semantic re-ranking strategies optimized for multi-lingual texts, shlokas, and Sanskrit sandhi/compounds.
- →Build rigorous hallucination-detection and citation-verification pipelines using LLM-as-a-Judge and ground-truth validation datasets.
- →Fine-tune open-source models (Llama, Mistral, Gemma, IndicLLM) for high-fidelity domain-specific comprehension and translation fidelity.
- →Optimize inference latencies, streaming response protocols, and token economics across production workloads.
- →Collaborate closely with backend engineers and Sanskrit researchers to continually expand our verified knowledge ontologies.
04Qualifications & Required Skills
Required Qualifications
- ✓2+ years of professional hands-on experience building production RAG systems, vector search pipelines, and LLM applications.
- ✓Strong proficiency in Python, modern AI frameworks (LangChain, LlamaIndex, LiteLLM), and vector databases (Qdrant, pgvector, or Milvus).
- ✓Solid understanding of embeddings, semantic similarity, re-ranking algorithms (Cohere, Cross-Encoders), and prompt engineering.
- ✓Experience deploying models and microservices with Docker, FastAPI, and cloud platforms (AWS, GCP, or Azure).
- ✓Sound understanding of data structures, search mechanics, and evaluation metrics (RAGAS, BLEU, ROUGE, Precision@K).
Key Skills & Technologies
Preferred / Desirable Skills
- ★Familiarity with Indic NLP, Devanagari script processing, or transliteration standards (IAST, Harvard-Kyoto, SLP1).
- ★Experience with Knowledge Graphs (Neo4j, RDF, GraphRAG architectures).
- ★Contributions to open-source AI libraries or published research in NLP/LLMs.
- ★Understanding of low-latency streaming interfaces and modern Next.js client-side streaming.
05Candidate Privacy Notice (DPDP Act, 2023)
Transparent Personal Data Processing:
Candidate personal data provided during application is collected and processed solely for recruitment evaluation, role matching, and communication in compliance with the Digital Personal Data Protection Act, 2023.
We do not collect unnecessary sensitive data (such as Aadhaar, PAN, or financial details) during initial applications. You may request correction or deletion of your candidate record at any time by contacting [email protected].
06Frequently Asked Questions
Q: Is this role remote?
Yes, this role is fully remote. Team members collaborate asynchronously across flexible time zones with core meetings aligned with Indian Standard Time.
Q: What is SattvaLok's tech stack for AI and RAG?
Our stack utilizes Python, FastAPI, LangChain/LlamaIndex, Qdrant/pgvector, hybrid BM25 + dense re-ranking, and high-performance Next.js and Kotlin Spring Boot services.
Q: How do you handle Sanskrit and Indic text tokenization?
We engineer custom pre-processing and transliteration tokenizers (handling IAST, Devanagari, and sandhi splitting) and evaluate specialized Indic embedding models to ensure accurate semantic similarity.
Q: What is the technical evaluation process?
The process includes an exploratory introductory call, an architecture deep-dive where you present past RAG engineering solutions, and a collaborative system design discussion.
Q: Can I join on a contract or part-time basis?
We welcome both high-commitment full-time staff engineers and specialized consultants/contractors who bring exceptional expertise in RAG and vector retrieval.
Stay Connected For Future Opportunities
Hiring for AI & RAG Engineer — Knowledge Systems is currently closed. Introduce yourself to be considered when matching opportunities open.
Submit General Expression of Interest