y
yash_fivacc

Yash Jagtap

@yash_fivacc

AI Software Engineer

Índia
Inglês, Hindi, Marata
Algumas informações são exibidas no idioma inglês.
Sobre mim
Hi! I'm an AI Software Engineer and currently an Associate AI Engineer at Dataspan Technologies, building AI/ML solutions that solve real business problems. I've automated data pipelines (60% less manual effort), built RAG-based LLMs with LangChain, developed predictive ML and fraud detection models (92% accuracy), and created 10+ BI dashboards. I'm looking for a team where I can build impactful AI products, solve challenging problems, and deliver measurable results end-to-end.... Saiba mais

Habilidades

y
yash_fivacc
Yash Jagtap
offline • 

Conheça meus serviços

Software e Sites de IA
I will build custom ai agents, rag pipelines, and llm automation solutions
A partir deUS$ 50 /projetoMais detalhes
Sites Personalizados
I will build modern web applications using python, fastapi, and react
A partir deUS$ 80 /projetoMais detalhes

Portfólio

Experiência profissional

DATA

Associate Software Developer

DATA • Período integral

Jan 2025 - Present • 1 yr 7 mos

▸ Developed multiple AI and forecast solutions that aligned with client requirements. ▸ Reduced unreviewed AI output errors by 40% by designing an end-to-end LLM automation pipeline (LangChain + LangGraph + OpenAI API) on 100K+ enterprise records, with LLM-as-Judge scoring and human-in-the-loop approval gates. ▸ Cut manual analyst processing time by 60% by building multi-agent agentic workflows with tool calling, memory management, state orchestration, confidence-threshold escalation, and n8n event-driven automation. ▸ Improved context-relevant RAG accuracy by engineering a full chunking → embedding → ChromaDB retrieval pipeline with retrieval tuning, FastAPI serving, response validation, and structured logging — shipped across 3 Agile sprint cycles. ▸ Surfaced pipeline metrics to business stakeholders via 10+ Power BI dashboards backed by SQL; monitored production LLM systems for drift, hallucination, and token cost — maintaining quality regression below threshold.