Principal Data Scientist @ Eximius AI | Architecting multi-agent AI systems, RAG platforms, and intelligent automation for modern recruitment

📍 Noida, India  |  💼 LinkedIn

I'm a Principal Data Scientist building Eximius — an AI-powered recruitment platform that replaces resume-driven guesswork with structured, explainable, job-relevant screening. I design and ship production-grade multi-agent AI systems, RAG pipelines, and LLM-powered automation across the entire hiring lifecycle — from job ingestion to candidate sourcing, parsing, screening, and interviewing.

• 🔬 Currently building: Multi-agent AI screening systems, RAG-based enterprise assistants, and document intelligence pipelines
• 🧠 Core focus: LLMs, LangGraph agent orchestration, RAG, NLP, and evaluation/observability for production AI
• 🎯 Interested in: Explainable AI decisions, human-in-the-loop systems, and scalable AI-native platforms
• 💬 Ask me about: Multi-agent architectures, RAG, LLM evaluation, NLP pipelines, ATS/API integrations

🤖 AI Screening Interviewer — Multi-Agent Screening & Assessment Architected a multi-agent platform (Python, LangGraph, LLMs, RAG) with specialized Resume, Job, Screening, Interview, and Assessment agents that generate evidence-based, explainable candidate assessments. Built an adaptive AI interviewer that dynamically generates follow-up questions based on competency coverage and confidence — with a human-in-the-loop framework so AI surfaces evidence, people decide.

🌐 Sourcing Cron — Multi-Source Candidate Sourcing & Enrichment Built a scalable sourcing platform unifying Dice, Monster, CareerBuilder, Nexxt, and Apify behind a single abstraction layer. Automated JD-driven candidate discovery, enrichment, and deduplication with production-grade pipelines (throttling, retries, idempotency) feeding into semantic candidate–job matching and ranking.

📄 Document Intelligence Platform — Resume/JD Parsing Architected a hybrid NLP + LLM pipeline (NER, semantic matching, schema-constrained extraction, OCR) converting heterogeneous resumes and JDs into standardized, validated JSON. Built canonical Resume & Job Intelligence schemas powering downstream matching, semantic search, and RAG-based recruiter intelligence.

🔗 Multi-ATS Recruitment Orchestration Platform Designed bi-directional ATS integrations (API/webhook/email) with an event-driven pipeline: Job Ingestion → Resume Parsing → Matching → AI/Rule-based Screening → Qualification → sync back to ATS. Built on Python, Azure Functions, event-driven queues, and Cosmos DB/PostgreSQL with full retry/idempotency handling.

💬 Enterprise Knowledge RAG Assistant Built a conversational RAG assistant for querying private company knowledge (HR policies, SOPs, wikis) with intelligent query routing between RAG retrieval and general LLM reasoning. Implemented hybrid vector/keyword search, reranking, source-grounded citations, RBAC, and hallucination safeguards for secure enterprise use.

Across all of these: LLM evaluation & observability — groundedness, extraction accuracy, hallucination rate, latency, cost, and reliability — is built in from day one, not bolted on after.

AI / LLM Systems

Languages & Core

Data & ML

Infra & Cloud

• 🦈 Pull Shark
• ❄️ Arctic Code Vault Contributor

Thanks for stopping by — feel free to explore my repositories and reach out!