I'm an AI & Data Engineer based in India, with over 4 years of experience bridging large-scale data engineering with production AI systems. I work across Python (PySpark), SQL, Databricks, and cloud-native architectures on Azure / GCP / AWS, with a focus on data quality and real-time pipelines.

I've delivered onsite with ministry stakeholders in Dubai and Abu Dhabi, translating government-level requirements into production data systems across UAE / UK / India time zones.

name:        Yashi Mishra
role:        AI & Data Engineer
location:    India  →  delivering globally (UAE • UK • IN)
focus:       Data Quality • Real-Time Pipelines • LLM & Voice AI
education:   B.Tech, Computer Science — GLA University (81.9%)
currently:   Researching data · Creating architectures · Medallion pipelines
ask-me-about: Microsoft Fabric, Databricks, PySpark, RAG, Whisper/Deepgram/ElevenLabs

• 🏛️ Authored 300+ data-quality rules for UAE Official Development Assistance pipelines on Microsoft Fabric + Azure Databricks — audit-grade ministry reporting with Unity Catalog lineage.
• ⚡ Cut SQL latency 30%+ to under 2 seconds for live-ops dashboards by rewriting queries and stored procedures across PostgreSQL + BigQuery.
• 🧠 Fine-tuned multilingual LLMs on the Hugging Face stack with parallel and batch training pipelines on distributed Azure + Ray infrastructure.
• 🎙️ Prototyped a real-time Voice AI interview assistant — Whisper + Deepgram + ElevenLabs — with full production architecture and sizing presented to the client.
• ❄️ Built a Bronze → Diamond medallion architecture on GCP for a cold-chain IoT logistics client, with N8N / Make.com alerting on reefer-temperature thresholds.
• 💼 Designed the backend data architecture for ZipApply's candidate-recruiter matching using a life-quality scoring signal, HPCC ECL ETL pipelines, and OpenAI-powered resume tooling.

AI / Voice / LLM Systems

Data Engineering

Backend & Languages

Cloud, DevOps & Viz

• 🎙️ Production architectures for low-latency Voice AI (sub-300ms end-to-end)
• 🧪 Distributed fine-tuning patterns for multilingual LLMs on Ray + Azure
• 🏗️ Lakehouse design with Unity Catalog and lineage-first data governance
• 🔄 Event-driven medallion pipelines that hold up under audit

"Data that ships, AI that talks, pipelines that hold under audit."