Aman Kaushik

I’m a Python Backend & AI Engineer with 1+ year of production experience designing and deploying Generative AI systems, multi-agent workflows, and scalable backend architectures.

At Qolaba AI, I engineered a production-scale knowledge graph and LLM pipeline (Neo4j) to automate OCR-based content structuring, solving data extraction bottlenecks for NCERT and state board curricula. I also built production-grade document classification and multilingual translation pipelines for the Maharashtra Government (Tapaal), processing thousands of daily inquiries across regional vernaculars.

I orchestrate agentic multi-agent workflows for Fortune 500 enterprise automation, leveraging advanced state management and agentic frameworks for high-reliability, mission-critical operations. I’ve also architected a provider-agnostic multi-modal engine for video, audio, and image synthesis, and built a polyglot backend (Python/TypeScript) designed for horizontal scalability and feature parity across APIs.

Across 5+ production client projects, I’ve deployed observability and evaluation pipelines using LangSmith and Opik to track hallucination rates, classification precision, and translation quality. Previously, as a Machine Learning Intern, I fine-tuned YOLOv8 models (0.95 mAP) and built FastAPI services for liveness detection and document processing (98% classification precision).