Open to opportunities

Aryan Kale

@aryankale

Message

I build production-ready LLM systems with RAG and scalable backends, optimizing latency, cost, and response quality.

India

Message

What I'm looking for

I’m looking to build and deploy LLM-powered products—owning RAG pipelines, evaluation, and backend reliability while balancing latency, cost, and answer quality in a collaborative team.

I’m an engineer focused on building production-ready LLM systems using RAG, multi-step reasoning pipelines, and scalable backend architectures. I care about real deployment constraints—latency, cost, and response quality—so the systems I build perform reliably in production.

I’ve designed retrieval pipelines, evaluation workflows, and AI-powered applications with end-to-end delivery in mind. In my recent experience, I also managed project workflows and coordinated cross-functional data to ensure client delivery compliance and milestone tracking.

In my projects, I’ve built systems like PaperMind, where I architected a multi-tenant RAG pipeline using a 13-step autonomous agent system for query planning, multi-hop reasoning, and document synthesis. I also engineered scalable vector storage with PostgreSQL and pgvector, migrating from local FAISS indices to persistent, isolated user data.

I prioritize quality control in the generation loop—implementing self-reflection and an evaluation agent to verify citation grounded-ness and halt hallucinations before answer generation. From HireReady (AI resume analysis with ATS scoring) to NewsPulse AI (an autonomous Telegram news bot with caching and deduplication), I focus on building practical, efficient AI products that handle edge cases like API rate limits and strict messaging constraints.

Experience

Work history, roles, and key accomplishments

PaperMind Deep Research Assistant

PaperMind

Feb 2026 - Mar 2026 (1 month)

Architected a multi-tenant, multi-agent RAG pipeline using a 13-step autonomous system for complex query planning, multi-hop reasoning, and document synthesis. Engineered hybrid retrieval and an evaluation self-reflection loop to improve citation grounded-ness and reduce hallucinations before answer generation.

RAG Multi Agent Systems fastAPI Faiss Gemini API Hybrid Retrieval (ANN + BM25)

Analyst (PMO & Operations)

Capgemini

Oct 2025 - Jan 2026 (3 months)

Managed project workflows and coordinated enterprise client operations for Aptiv (Americas). Streamlined reporting and analyzed operational data to support milestone tracking and resource allocation.

PMO Reporting Resource Allocation

HireReady AI Resume Analyzer

HireReady

Oct 2025 - Nov 2025 (1 month)

Built an AI resume analysis platform with ATS scoring and job-description matching to generate structured feedback on keyword gaps and improvement areas. Implemented JWT authentication and secure password hashing with bcrypt and stored user resume history.

React Flask Gemini API JWT Bcrypt MongoDB MySQL

NewsPulse AI News Aggregation Bot

NewsPulse AI

Mar 2026 - Present (3 months)

Built an autonomous Telegram news bot using Gemini 2.5 Flash and Google Search grounding to curate and broadcast real-time briefings across 10 topics. Implemented sequential caching and dynamic chunking to bypass API rate limits and Telegram message length constraints, with deployment on Railway and deduplication.

Python Gemini API Telegram Bot API AsyncIO Google Search Grounding Caching SQLite